diff --git a/phaseR/behavior-1k/boxing_food_after_dinner/trajectories/behavior-boxing-food-after-dinner-i00-standard--vpambfm.json b/phaseR/behavior-1k/boxing_food_after_dinner/trajectories/behavior-boxing-food-after-dinner-i00-standard--vpambfm.json
new file mode 100644
index 0000000000000000000000000000000000000000..190c5ce92e9226f9a98bf0dbf65004ee7f954b0c
--- /dev/null
+++ b/phaseR/behavior-1k/boxing_food_after_dinner/trajectories/behavior-boxing-food-after-dinner-i00-standard--vpambfm.json
@@ -0,0 +1,420 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-boxing-food-after-dinner-i00-standard--vpambfm",
+ "taskKey": "behavior-1k/boxing_food_after_dinner",
+ "dag_sha256": "e8d426fd30e04edebad1d25c44ec0cecb6d3f45ff17585b7f91cdb42b93d8ef7",
+ "outcome": {
+  "success": 0,
+  "reward": 0.5,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 29287,
+   "sim_steps": 29287,
+   "physics_steps": 117148,
+   "steps": 15342,
+   "elapsed_s": 11975.0,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "replay_error": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "q_score": 0.5,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.334,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.459,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-boxing-food-after-dinner-i00-standard--vpambfm/replay.mp4",
+  "duration_s": 976.234,
+  "fps": 30.0,
+  "frames": 29287,
+  "sha256": "467eaed25731a40270c2790dbfa94d597d82dae6f43934882c12acfa58a19c51"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 8,
+  "base": 1,
+  "peak": 6,
+  "final": 6,
+  "nodes": {
+   "cabinet_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      288,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 288,
+      "log_step": null,
+      "text_zh": "观察：视频约 288-330 s 头部相机可见水槽下方橱柜门已离开柜体，柜内暴露；精确开门时刻约±10 s（仅粗采样）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓握/拉开橱柜把手（推断）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动到相应位置支撑机械臂"
+     }
+    ]
+   },
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      752,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 755,
+      "log_step": null,
+      "text_zh": "观察：视频 600-740 s 冰箱门仍关，750 s 夹爪正拉门把手，755 s 门已敞开露出搁板；开门约 750-755 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪拉开冰箱门把手（推断）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动到相应位置支撑机械臂"
+     }
+    ]
+   },
+   "tacos_in_box_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      540,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 556,
+      "log_step": null,
+      "text_zh": "观察：视频 556 s 左右保鲜盒被夹爪托住，盒内可见塔可；第一个塔可入盒时间约 500-540 s，粗略。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓握塔可放入盒内"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动到相应位置支撑机械臂"
+     }
+    ]
+   },
+   "tacos_in_box_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      675,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 680,
+      "log_step": null,
+      "text_zh": "观察：视频 600、620 s 台面盘中仍留有一个塔可，660 s 夹爪仍持塔可，680 s 腕部相机见盒内两个塔可，约 660-690 s 第二个入盒；随后盒被送入冰箱。q_score=0.5（评分器字段）与 塔可×2+盒入冰箱 共3项一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓握塔可放入盒内"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动到相应位置支撑机械臂"
+     }
+    ]
+   },
+   "box_in_fridge": {
+    "method": "visual",
+    "intervals": [
+     [
+      880,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 900,
+      "log_step": null,
+      "text_zh": "观察：视频 800 s 持盒在台面，820-880 s 送入冰箱，900 s 盒在冰箱搁板上，夹爪多次把盒往里推；最终位置在冰箱内。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪托持/抓握保鲜盒送入冰箱后用手推到深处"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驶向冰箱并支撑机械臂"
+     }
+    ]
+   },
+   "plates_on_sink_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 975,
+      "log_step": null,
+      "text_zh": "观察：视频开头与结尾（975 s）盘子都在台面上，没有任何盘子进入水槽；全程画面未见机器人接触盘子。"
+     }
+    ]
+   },
+   "plates_on_sink_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 975,
+      "log_step": null,
+      "text_zh": "观察：同上，两盘都未入水槽。"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      752
+     ],
+     [
+      972,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：冰箱初始关闭，视频 0-740 s 门关着，约 752 s 开门后条件失去；S4：t=0 成立的状态从 0 起计。"
+     },
+     {
+      "t": 975,
+      "log_step": null,
+      "text_zh": "观察：视频 955-975 s 机器人用手臂推冰箱门，975 s 画面中白色门板已贴回箱体（粗略判断，终帧被门遮挡）；log step 364-367 为此推门动作。推门后恢复；q 总分不计初始即成立项，与最终值不矛盾。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "夹爪/手臂推冰箱门至关闭"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动到冰箱旁"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.5,
+   "unit": "1",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "plates_on_sink_1",
+   "where": "首步前",
+   "t": 975,
+   "log_step": null,
+   "evidence_zh": "观察：视频全程两个盘子留在台面，结尾 975 s 仍在台面上；机器人把塔可、保鲜盒、关门都做了，从未处理盘子。log 最后一步 368 为 API 报 token 预算用尽（run 结束）。"
+  },
+  "phenomena": [
+   {
+    "t": 975,
+    "log_step": null,
+    "evidence_zh": "观察：视频中未见夹爪接触任何盘子，盘子未入水槽（0/2）。预算用尽不作为现象。",
+    "cat": "没尝试",
+    "item": "没尝试"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "根因不明：无状态日志；只见预算耗尽（log step 368）而盘子步骤从未开始，没有证据说明为何先后顺序如此或计划错误，故根因留空。节点时间仅粗采样，约±10-20 s。",
+  "unmet": {
+   "tacos_in_box_1": true,
+   "tacos_in_box_2": true,
+   "box_in_fridge": true,
+   "plates_on_sink_1": false,
+   "plates_on_sink_2": false,
+   "fridge_closed": true
+  }
+ },
+ "summary_zh": "观察：视频 976 s 内机器人打开橱柜、取保鲜盒装入两个塔可、打开冰箱放入盒子并推门关上（约 6/8 节点，q_score 0.5），两只盘子始终留在台面，未入水槽。断点为盘子入水槽（从未开始），在 log step 368 因 token 预算用尽结束；节点时间为粗略视觉估计，无状态日志。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 3,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "总体（6/8、断点在盘子步骤、没尝试、根因留空）与视频一致；唯一需改：fridge_closed 是 state 节点，冰箱开局即关闭，应从视频 0 s 起成立，开门后失去，972 s 再恢复（S4，不受 after 限制）。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "总体（6/8、断点在盘子、没尝试）成立，但中段时间错误：fridge_opened/fridge_closed 的开门时刻被写成约 590 s，实际视频 750-755 s 才开门；tacos_in_box_2 写成 556 s，实际 600/620 s 台面盘中仍有一个塔可，第二个入盒约 660-680 s。",
+    "n_issues": 3,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 3,
+    "kind": "confirm",
+    "verdict": "agree",
+    "summary_zh": "三项 round-2 问题均已按审稿意见正确落实，与自查帧一致，未发现新的事实错误。",
+    "n_issues": 0,
+    "n_checks": 3
+   }
+  ],
+  "open_zh": [],
+  "confirm": {
+   "verdict": "confirmed",
+   "summary_zh": "三项 round-2 问题均已按审稿意见正确落实，与自查帧一致，未发现新的事实错误。",
+   "checks_ok": "3/3",
+   "n_factual_errors": 0
+  }
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T21:00:15Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-boxing-food-after-dinner-i00-standard--vpambfm/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-boxing-food-after-dinner-i00-standard--vpambfm/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-boxing-food-after-dinner-i00-standard--vpambfm/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-boxing-food-after-dinner-i00-standard--vpambfm/04_review_r2"
+   },
+   {
+    "stage": "reconcile",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-boxing-food-after-dinner-i00-standard--vpambfm/05_reconcile_r2"
+   },
+   {
+    "stage": "confirm",
+    "round": 3,
+    "ok": true,
+    "space": "work/runs/behavior-boxing-food-after-dinner-i00-standard--vpambfm/06_confirm_r3"
+   }
+  ],
+  "confirm_prompt": "prompts/06_confirm.md",
+  "confirm_prompt_sha256": "b9f56c095f7ea5e609e58bb21e076f565766c370333d71c7a3512cffb4d78267",
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6188c6146807cec6439472bb42defc5d9562b7cc383f831205b23bd5811b4c4c",
+   "video_sha256": "467eaed25731a40270c2790dbfa94d597d82dae6f43934882c12acfa58a19c51",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/bringing_in_wood/trajectories/behavior-bringing-in-wood-i00-standard--tuynadm.json b/phaseR/behavior-1k/bringing_in_wood/trajectories/behavior-bringing-in-wood-i00-standard--tuynadm.json
new file mode 100644
index 0000000000000000000000000000000000000000..81c88b76cb99f5e89d6954b1ab54bf86d165f571
--- /dev/null
+++ b/phaseR/behavior-1k/bringing_in_wood/trajectories/behavior-bringing-in-wood-i00-standard--tuynadm.json
@@ -0,0 +1,264 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-bringing-in-wood-i00-standard--tuynadm",
+ "taskKey": "behavior-1k/bringing_in_wood",
+ "dag_sha256": "54359bd2d9d51967b98ef078e09c231542698021d39d112e7af391cea06ee0e7",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 6543,
+   "sim_steps": 6693,
+   "physics_steps": 26772,
+   "steps": 1462,
+   "elapsed_s": 6234.7,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 150,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.0,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.366,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.29,
+   "peak_cpus": 1.18,
+   "peak_pids": 151,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-bringing-in-wood-i00-standard--tuynadm.mp4",
+  "duration_s": 223.2,
+  "fps": 10.0,
+  "frames": 2232,
+  "sha256": "cd72c9d37701b6cb0e7f5212da85dd8f0172715ed13af14af0f5b91ef54dbad5"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 4,
+  "base": null,
+  "peak": null,
+  "final": 1,
+  "nodes": {
+   "corridor_door_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      93.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 95.0,
+      "log_step": null,
+      "text_zh": "观察：视频 80 s 头部相机中玻璃木门仍关闭，机器人右夹爪在把手旁；视频 95 s 门扇已转开，110-215 s 一直可见门扇开启、内侧走廊木地板。推断：开门约在视频 90-95 s 完成（时间不精确）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右夹爪握住门把手牵引门扇（视频 80-95 s 可见夹爪在把手处，门随后转开）；底盘移动配合"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "轮式底盘移动配合开门"
+     }
+    ]
+   },
+   "plywoods_on_corridor_1": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 215.0,
+      "log_step": null,
+      "text_zh": "观察：视频 100-220 s 左腕相机可见一块橙色板被夹在两指之间（150 s、200 s 整帧可见），头部相机见机器人仍在花园侧门口，板未放到走廊木地板；q_score=0.0。推断：板始终未入走廊，未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "左夹爪夹持橙色板（视频约 40-60 s 与 100-220 s 左腕相机可见）；板未送入走廊"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "轮式底盘移动携板（推断）"
+     }
+    ]
+   },
+   "plywoods_on_corridor_2": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 215.0,
+      "log_step": null,
+      "text_zh": "观察：视频 215 s 头部相机见板未入走廊；评分器 q_score=0.0。推断：未达成。"
+     }
+    ]
+   },
+   "plywoods_on_corridor_3": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 215.0,
+      "log_step": null,
+      "text_zh": "观察：视频 215 s 头部相机见板未入走廊；评分器 q_score=0.0。推断：未达成。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "比例",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "plywoods_on_corridor_1",
+   "where": "操作中",
+   "t": 150.0,
+   "evidence_zh": "观察：门于视频约 95 s 开启后，视频 100-220 s 左腕相机见一块板被夹持，机器人仍在花园侧门口，未把板送入走廊；q_score=0。",
+   "log_step": null
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "视频 100-220 s 板被夹持却长时间停在门口未进走廊，断点在携板通过/放置而非抓取；但三路相机视角遮挡且无状态日志，无法判定是被门挡、路径未走完还是其他原因，因此不给根因。agent 日志在 6000+ s 仍反复调试开门（log step 198-219），属 session 证据。",
+  "unmet": {
+   "plywoods_on_corridor_1": false,
+   "plywoods_on_corridor_2": false,
+   "plywoods_on_corridor_3": false
+  }
+ },
+ "summary_zh": "观察：图内开了走廊门（视频约 95 s），左夹爪在 100-220 s 夹持一块板但始终停在门口，未入走廊，q_score=0，0/3 板入走廊。断点在携板进走廊；无状态日志且视角遮挡，根因留空。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "门开启（约95 s）判断可接受，但标签称“全程没有板被夹起”与视频矛盾：左腕相机在约100-220 s（视频时钟）持续显示橙色木板夹在夹爪指间，约40-60 s亦有。板未入走廊，故未达成成立，但 plywood_1 的证据和 contact_actual、断点证据以及 gap 需改写。",
+    "n_issues": 4,
+    "n_major": 3
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "门开启（视频约95 s）、板被左夹爪夹持但始终未入走廊（视频100-220 s）、q_score=0 与帧一致；断点 plywoods_on_corridor_1 合理，空的 phenomena/root_causes 带有 gap_zh，证据不足以支持更具体的原因。未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:42Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-bringing-in-wood-i00-standard--tuynadm/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-bringing-in-wood-i00-standard--tuynadm/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-bringing-in-wood-i00-standard--tuynadm/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-bringing-in-wood-i00-standard--tuynadm/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "32ea39526507b0ab54d7ec80e1d105e033edf44c2416b833053a7e7e65307330",
+   "video_sha256": "cd72c9d37701b6cb0e7f5212da85dd8f0172715ed13af14af0f5b91ef54dbad5",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/can_meat/trajectories/behavior-can-meat-i00-standard--pdtpfwh.json b/phaseR/behavior-1k/can_meat/trajectories/behavior-can-meat-i00-standard--pdtpfwh.json
new file mode 100644
index 0000000000000000000000000000000000000000..199af4b37b128bcd077c83cbcea163aef0d4c2df
--- /dev/null
+++ b/phaseR/behavior-1k/can_meat/trajectories/behavior-can-meat-i00-standard--pdtpfwh.json
@@ -0,0 +1,342 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-can-meat-i00-standard--pdtpfwh",
+ "taskKey": "behavior-1k/can_meat",
+ "dag_sha256": "f3528823deacaae41ee48f35af2dbb099ba817ce76d5ff2cd5f1d48523bc2056",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 22040,
+   "sim_steps": 22190,
+   "physics_steps": 88760,
+   "steps": 1585,
+   "elapsed_s": 6686.1,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 150,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.0,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "peak_cpus": 0.0,
+   "peak_pids": 0.0,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-can-meat-i00-standard--pdtpfwh.mp4",
+  "duration_s": 739.7,
+  "fps": 10.0,
+  "frames": 7397,
+  "sha256": "bea780751c9eebe24ba2db5208de2388027e864dd64c4e1e5ca4d5285064014d"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 12,
+  "base": 4,
+  "peak": 4,
+  "final": 4,
+  "nodes": {
+   "cabinet_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 739.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-739.7 s 的头部/腕部画面中，未见橱柜门或罐盖被打开，也未见香肠被夹取或入罐；评分器 q_score=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "agent 在 log 中多次指令右臂抓握并拉动把手（如 log step 788 水槽柜把手、log step 844 微波炉下方抽屉把手）；视频中未见目标橱柜门打开，具体接触不能完全确认。"
+     }
+    ]
+   },
+   "jars_opened_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 739.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-739.7 s 的头部/腕部画面中，未见橱柜门或罐盖被打开，也未见香肠被夹取或入罐；评分器 q_score=0。"
+     }
+    ]
+   },
+   "jars_opened_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 739.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-739.7 s 的头部/腕部画面中，未见橱柜门或罐盖被打开，也未见香肠被夹取或入罐；评分器 q_score=0。"
+     }
+    ]
+   },
+   "sausages_in_jars_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 739.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-739.7 s 的头部/腕部画面中，未见橱柜门或罐盖被打开，也未见香肠被夹取或入罐；评分器 q_score=0。"
+     }
+    ]
+   },
+   "sausages_in_jars_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 739.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-739.7 s 的头部/腕部画面中，未见橱柜门或罐盖被打开，也未见香肠被夹取或入罐；评分器 q_score=0。"
+     }
+    ]
+   },
+   "sausages_in_jars_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 739.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-739.7 s 的头部/腕部画面中，未见橱柜门或罐盖被打开，也未见香肠被夹取或入罐；评分器 q_score=0。"
+     }
+    ]
+   },
+   "sausages_in_jars_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 739.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-739.7 s 的头部/腕部画面中，未见橱柜门或罐盖被打开，也未见香肠被夹取或入罐；评分器 q_score=0。"
+     }
+    ]
+   },
+   "jars_closed_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：两罐起初关闭（DAG），视频全程未见罐盖被翻开；推断：一直保持关闭。"
+     }
+    ]
+   },
+   "jars_closed_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：两罐起初关闭（DAG），视频全程未见罐盖被翻开；推断：一直保持关闭。"
+     }
+    ]
+   },
+   "jars_in_cabinet_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "推断：罐起初在柜内（DAG），视频中柜门始终未打开、未见取出罐，故保持在柜内。"
+     }
+    ]
+   },
+   "jars_in_cabinet_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "推断：罐起初在柜内（DAG），视频中柜门始终未打开、未见取出罐，故保持在柜内。"
+     }
+    ]
+   },
+   "cabinet_closed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 739.0,
+      "log_step": null,
+      "text_zh": "推断：终态节点须在前置节点（打开橱柜、香肠入罐等）达成后才成立，这些均未达成，故为假。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "1",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cabinet_opened",
+   "t": null,
+   "log_step": 788,
+   "where": "探索",
+   "evidence_zh": "观察：视频 0-739.7 s 采样帧（约每 100 s 一帧及密集抽查）未见任何橱柜门打开；log step 788、844 显示 agent 在 agent 工作第 6000 s 以后仍在拉水槽柜把手和微波炉下方抽屉把手，q_score=0，从未完成第一步。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "视频是多相机拼接且画面大多被机身/台面遮挡，无状态日志；无法确认具体是没碰到把手还是拉不动，也无法证明 agent 找错了橱柜，故现象与根因留空。",
+  "unmet": {
+   "sausages_in_jars_1": false,
+   "sausages_in_jars_2": false,
+   "sausages_in_jars_3": false,
+   "sausages_in_jars_4": false,
+   "jars_closed_1": true,
+   "jars_closed_2": true,
+   "jars_in_cabinet_1": true,
+   "jars_in_cabinet_2": true,
+   "cabinet_closed": false
+  }
+ },
+ "summary_zh": "观察：视频 739.7 s 内未见橱柜被打开，任何香肠都没有被拿取，q_score=0。只有起初就成立的罐关闭、罐在柜内节点保持为真，其余节点均未达成；断在第一步“打开橱柜”。agent 在 log 中反复尝试拉把手/抽屉，但视频不足以判定具体失败现象和根因。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "q_score=0 且视频采样帧未见橱柜打开；t=0 即成立的四个节点保持为真与 q 一致，断点 cabinet_opened、现象与根因留空并附 gap_zh 均有依据；未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:53:33Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-can-meat-i00-standard--pdtpfwh/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-can-meat-i00-standard--pdtpfwh/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ed2bcb6d9459e4719f26fce74f136452d6acaf08826ef8900a723c6f23193aba",
+   "video_sha256": "bea780751c9eebe24ba2db5208de2388027e864dd64c4e1e5ca4d5285064014d",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/canning_food/trajectories/behavior-canning-food-i00-standard--xz6g2se.json b/phaseR/behavior-1k/canning_food/trajectories/behavior-canning-food-i00-standard--xz6g2se.json
new file mode 100644
index 0000000000000000000000000000000000000000..eea2c05d416dbe5b07e77fa42da86a2433f51b4c
--- /dev/null
+++ b/phaseR/behavior-1k/canning_food/trajectories/behavior-canning-food-i00-standard--xz6g2se.json
@@ -0,0 +1,427 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-canning-food-i00-standard--xz6g2se",
+ "taskKey": "behavior-1k/canning_food",
+ "dag_sha256": "79f4ae898ca53c66f1426df6e0cf977ab188a121892679dbdd743d829b8da565",
+ "outcome": {
+  "success": 0,
+  "reward": 0.2,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 12768,
+   "sim_steps": 12918,
+   "physics_steps": 51672,
+   "steps": 3940,
+   "elapsed_s": 6614.1,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 150,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.2,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "peak_cpus": 0.0,
+   "peak_pids": 0.0,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-canning-food-i00-standard--xz6g2se.mp4",
+  "duration_s": 430.7,
+  "fps": 10.0,
+  "frames": 4307,
+  "sha256": "137c96457077b9d8749f8ebcae0767035babc0a0e1824a4f9e993607c8d4b335"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 10,
+  "base": 4,
+  "peak": 4,
+  "final": 3,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      202,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 202,
+      "log_step": null,
+      "text_zh": "观察：视频 200–204 s 头部视角冰箱门板开始离开闭合位，212 s 后大开露出搁板，腕部视角见菠萝在搁板上（约±8 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓握门把手拉开门；底盘移动支撑"
+     }
+    ]
+   },
+   "cabinet_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      24,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 24,
+      "log_step": null,
+      "text_zh": "观察：视频约 24–40 s 头部视角可见橱柜门板打开、内腔露出（时间为8 s间隔抽帧估计，约±8 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓握门把手拉开门；底盘移动支撑"
+     }
+    ]
+   },
+   "steak_diced": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 335,
+      "log_step": 390,
+      "text_zh": "观察：视频约 335–366 s 夹爪多次尝试从冰箱下层搁板取牛排未成功，未见牛排被取出或切；agent 自述（log step 390–429、470）仅作参考。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪在冰箱搁板上尝试抓牛排，抓握未成功"
+     }
+    ]
+   },
+   "pineapple_diced": {
+    "method": "visual",
+    "intervals": [
+     [
+      400,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 400,
+      "log_step": null,
+      "text_zh": "观察：视频约 400 s 后腕部视角可见刀具与菠萝旁出现黄色丁粒，约 410 s 起碗旁可见散落菠萝丁（约±8 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "压/擦",
+      "note_zh": "夹爪持刀以刀刃切菠萝"
+     }
+    ]
+   },
+   "steak_bowl_filled": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：牛排未取出，无牛排丁，碗中无牛排丁。"
+     }
+    ]
+   },
+   "pineapple_bowl_filled": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：视频末帧碗倾倒，菠萝丁洒在岛台上；碗内是否超过20%无法确认，按画面判未达成（最后一帧碗翻倒，推断：不满足）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓碗并移动，碗倾倒洒料"
+     }
+    ]
+   },
+   "bowls_in_cabinet_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      180
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：初始两碗在柜内；视频 139–146 s 岛台仅一只碗，约 178–184 s 夹爪抓着第二只碗离开橱柜，190–193 s 岛台上两只碗，之后碗一直在柜外直到结束（约±5 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓碗取出放到岛台；底盘移动"
+     }
+    ]
+   },
+   "bowls_in_cabinet_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      128
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：初始两碗在柜内；约 128 s 起第一只碗被取出，柜内不足2只。"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      202
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：初始冰箱闭合；约 202 s 被打开后直到视频结束未见关上。"
+     }
+    ]
+   },
+   "cabinet_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      24
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：初始橱柜闭合；约 24 s 打开后未见关上（视频中期仍开着）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "steak_diced",
+   "t": 335,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 335–366 s 夹爪伸入冰箱下层搁板处多次接近牛排，牛排未被取出（头/腕部视角可见搁板与夹爪）；log step 390–429 agent 自述多次抓取落空（仅作参考）。推断：steak_diced 的取牛排前置动作失败后改去切菠萝。",
+   "log_step": 390
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 335,
+    "log_step": 398,
+    "evidence_zh": "观察：视频约 335–366 s 夹爪在冰箱下层搁板牛排处多次动作，牛排始终留在搁板上、未被取出；log step 398/406 agent 自述手指闭合在牛排上方、牛排滑向搁板边缘（仅作参考）。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "evidence_zh": "观察：视频 335–366 s 牛排扁平贴在玻璃搁板上，多次同类抓取均未取出；log step 398/406/413 自述同类自上而下/斜向抓法反复失败（仅参考）；换用贴搁板侧向/铲取是否可行未验证。",
+    "source_zh": "视频"
+   }
+  ],
+  "gap_zh": "无状态日志，视频多处被遮挡；抓取失败的具体机理（几何/定位）只能由视频与 agent 自述推断，故根因置信度较低，替代抓法是否可行未验证。碗倾倒的自述（log step 470/475）不作为根因。",
+  "unmet": {
+   "steak_diced": false,
+   "pineapple_diced": true,
+   "steak_bowl_filled": false,
+   "pineapple_bowl_filled": false,
+   "bowls_in_cabinet_1": false,
+   "bowls_in_cabinet_2": false,
+   "fridge_closed": false,
+   "cabinet_closed": false
+  }
+ },
+ "summary_zh": "观察：机器人开了橱柜和冰箱，约 128–190 s 先后把两只碗取出放到岛台（碗入柜节点失去）；视频约 335–366 s 多次尝试从冰箱下层搁板取牛排未成功（没建立接触，操作中），之后改切菠萝，视频末帧碗倾倒洒出菠萝丁；门均未关，q_score 0.2。根因（抓取和接触设计）仅由视频与自述推断，置信度低。时间为抽帧估计。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 3,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点时间多数在±8 s估计内可接受，但 bowls_in_cabinet_1 的失去时间错：视频 129–146 s 只取出第一只碗，第二只碗约视频 178–190 s 才取出放到岛台，不是 136 s；另外牛排步骤从未开始，缺少「没尝试」现象，where 与 gap 需调整。",
+    "n_issues": 3,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "牛排步骤并非「没尝试」：log step 390–429 与视频约 330–368 s 都显示机器人多次伸手到冰箱下层搁板抓牛排但未取出；断点应为 steak_diced 操作中，现象应是没建立接触，而不是首步前/没尝试。节点时间基本可接受。",
+    "n_issues": 4,
+    "n_major": 3
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 3,
+    "kind": "confirm",
+    "verdict": "agree",
+    "summary_zh": "四个 accept 均已在 prev/trajectory.json 中完整落实，与视频 335–366 s 及 log step 390–429 一致，validate 通过。",
+    "n_issues": 0,
+    "n_checks": 4
+   }
+  ],
+  "open_zh": [],
+  "confirm": {
+   "verdict": "confirmed",
+   "summary_zh": "四个 accept 均已在 prev/trajectory.json 中完整落实，与视频 335–366 s 及 log step 390–429 一致，validate 通过。",
+   "checks_ok": "4/4",
+   "n_factual_errors": 0
+  }
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T21:00:25Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-canning-food-i00-standard--xz6g2se/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-canning-food-i00-standard--xz6g2se/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-canning-food-i00-standard--xz6g2se/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-canning-food-i00-standard--xz6g2se/04_review_r2"
+   },
+   {
+    "stage": "reconcile",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-canning-food-i00-standard--xz6g2se/05_reconcile_r2"
+   },
+   {
+    "stage": "confirm",
+    "round": 3,
+    "ok": true,
+    "space": "work/runs/behavior-canning-food-i00-standard--xz6g2se/06_confirm_r3"
+   }
+  ],
+  "confirm_prompt": "prompts/06_confirm.md",
+  "confirm_prompt_sha256": "b9f56c095f7ea5e609e58bb21e076f565766c370333d71c7a3512cffb4d78267",
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "feb1c652de82b90755141d694b5a14c9b89f9d23b059152cd8f01b7c8c928194",
+   "video_sha256": "137c96457077b9d8749f8ebcae0767035babc0a0e1824a4f9e993607c8d4b335",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/carrying_in_groceries/trajectories/behavior-carrying-in-groceries-i00-standard--v82xqwy.json b/phaseR/behavior-1k/carrying_in_groceries/trajectories/behavior-carrying-in-groceries-i00-standard--v82xqwy.json
new file mode 100644
index 0000000000000000000000000000000000000000..7c180c53b1614d1f0496825f730334288e3fb0eb
--- /dev/null
+++ b/phaseR/behavior-1k/carrying_in_groceries/trajectories/behavior-carrying-in-groceries-i00-standard--v82xqwy.json
@@ -0,0 +1,339 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-carrying-in-groceries-i00-standard--v82xqwy",
+ "taskKey": "behavior-1k/carrying_in_groceries",
+ "dag_sha256": "665ce9309b34a8b8f9c69c1468a50fd9eaf4d2059b6ec984d4cfd8806c04ff56",
+ "outcome": {
+  "success": 0,
+  "reward": 0.25,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 6823,
+   "sim_steps": 6823,
+   "physics_steps": 27292,
+   "steps": 1817,
+   "elapsed_s": 5179.9,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 0,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 1,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.25,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.269,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.283,
+   "peak_cpus": 0.89,
+   "peak_pids": 119,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-carrying-in-groceries-i00-standard--v82xqwy.mp4",
+  "duration_s": 227.5,
+  "fps": 10.0,
+  "frames": 2275,
+  "sha256": "ad95cb99c64d87761c029f1d6392c14a2b43d279f8fa86d4967693dc3bbee9ac"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 6,
+  "base": 1,
+  "peak": 3,
+  "final": 3,
+  "nodes": {
+   "garage_passage_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      54.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 54.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 48 s 机器人右夹爪握内侧木门把手，约 54 s 起木门已转开露出走廊并通过，之后视频中保持敞开（视频 223 s 返回时门仍开）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右夹爪握把手压下并推开木门；底盘驶近支撑"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驶到门旁"
+     }
+    ]
+   },
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      126.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 126.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 124-128 s 右夹爪拉/推冰箱门，门扇转开露出搁架内腔，此后保持敞开至 176 s 以后。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右夹爪握冰箱把手开门，随后推开"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移到冰箱旁"
+     }
+    ]
+   },
+   "tomato_in_fridge": {
+    "method": "visual",
+    "intervals": [
+     [
+      176.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 176.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 170-176 s 左夹爪持番茄伸入冰箱，松手后番茄留在中层搁架上，之后帧中仍在搁架内。q_score 0.25 与此一致（推断：1/4 目标已满足）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "左夹爪抓住番茄，经车库、走廊搬到冰箱放入"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "驾驶底盘搬运番茄到厨房"
+     }
+    ]
+   },
+   "milk_in_fridge": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 225.0,
+      "log_step": null,
+      "text_zh": "观察：全程只搬运番茄；牛奶盒留在车尾箱。机器人在视频约 190-227 s 返回车库方向，未再取牛奶（视频末尾停在走廊/门口，黑屏）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "驶回车库途中被门框卡住，未接触牛奶"
+     }
+    ]
+   },
+   "car_closed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 8.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 8-30 s 红色车后备厢盖抬起，露出黑色内腔，之后机器人从未回到车尾关闭；终段 213-227 s 在车库方向的帧中也未见关盖动作。"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0.0,
+      124.0
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频开局冰箱两门齐平关闭（t=0 已成立），约 124-126 s 被机器人拉开后至结束保持敞开；176-190 s 放入番茄后机器人离开冰箱未关门。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.25,
+   "unit": "fraction",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "milk_in_fridge",
+   "where": "转场",
+   "t": 200.0,
+   "evidence_zh": "观察：视频 176 s 放入番茄后机器人驶回车库取牛奶；log step 245 服务端输出显示穿过车库内门时 up_z 1.0→0.727→-0.0008（机体侧倒），step 260 回合以 'failure: the robot fell over' 结束；视频约 200-227 s 为走廊/门框处视角，末帧黑暗近墙，未到车尾，牛奶未取。推断（agent 自述）：旋转底盘 yaw 穿门时上身挂到门框。",
+   "log_step": 245
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "被挡/碰撞",
+    "t": 215.0,
+    "log_step": 245,
+    "evidence_zh": "观察：log step 245 服务端读数 up_z 降至 -0.0008（机体侧倒），step 260 'failure: the robot fell over'；视频约 215-227 s 画面倾斜并变暗贴近门/墙。推断（agent 自述）：上身挂到门框，视频无法清楚显示该接触。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "观察：服务端读数（log step 245/260）显示穿门时机体倾倒。推断（agent 自述）：侧向通过门洞时转动底盘 yaw 使上身挂在门框，该接触视频不可直接确认。",
+    "source_zh": "服务端读数（log step 245/260）+ agent 自述（挂框原因）",
+    "alt_feasible": "未知"
+   }
+  ],
+  "gap_zh": "无状态日志；倾倒由服务端读数证实，但'上身挂门框'的接触及 yaw 原因为 agent 自述推断，视频末段只有头部相机无法直接确认。未核实 car_closed 在 t=0 的初态（按视频敞开判为 false）。",
+  "unmet": {
+   "tomato_in_fridge": true,
+   "milk_in_fridge": false,
+   "car_closed": false,
+   "fridge_closed": false
+  }
+ },
+ "summary_zh": "观察：智能体打开车库内门、打开冰箱并把番茄放入冰箱（视频约 176 s），3/6 节点达成（q_score 0.25）。牛奶未搬、冰箱和汽车未关；返回车库穿门时机体倾倒（服务端读数 up_z→-0.0008，回合以 fell over 结束；挂框原因为 agent 自述），视频 227.5 s 结束。主要缺口：无状态日志，挂框接触依 agent 自述。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "已达成节点（开内门 ~54 s、开冰箱 ~126 s、番茄入冰箱 ~176 s）与 q=0.25 一致，断点 milk_in_fridge/转场也合理。但 fridge_closed 是开局即成立的 state 节点，标签给了空 intervals（应从 0 起），并且断点/现象的证据没有引用 log 中服务端对本回合的倾倒读数（step 245 up_z 0.727→-0.0008、step 260 'failure: the robot fell over'），仍标成'agent 自述'。",
+    "n_issues": 3,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "第2轮复核：节点（内门~54 s、冰箱~126 s、番茄~176 s、fridge_closed 从0起）与 q=0.25 一致；断点 milk_in_fridge/转场合理，视频 ~204-216 s 已到车库门口方向，log step 245 服务端读数 up_z 1.0→0.727(z=0.162)→-0.0008 证实倾倒。被挡/碰撞（z 抬升显示机体被门框顶起）与 控制 根因证据已标明推断部分，conf 5 诚实，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:13Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-carrying-in-groceries-i00-standard--v82xqwy/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-carrying-in-groceries-i00-standard--v82xqwy/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-carrying-in-groceries-i00-standard--v82xqwy/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-carrying-in-groceries-i00-standard--v82xqwy/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9387823c889101b4b7e398d7dfccc0dbd5a8aaf756098f5c3d87e9a9b8ad7caf",
+   "video_sha256": "ad95cb99c64d87761c029f1d6392c14a2b43d279f8fa86d4967693dc3bbee9ac",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/carrying_out_garden_furniture/trajectories/behavior-carrying-out-garden-furniture-i00-standard--xfwszou.json b/phaseR/behavior-1k/carrying_out_garden_furniture/trajectories/behavior-carrying-out-garden-furniture-i00-standard--xfwszou.json
new file mode 100644
index 0000000000000000000000000000000000000000..2ca90fbeff6b3b01365f89e97494369aa1b32308
--- /dev/null
+++ b/phaseR/behavior-1k/carrying_out_garden_furniture/trajectories/behavior-carrying-out-garden-furniture-i00-standard--xfwszou.json
@@ -0,0 +1,196 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-carrying-out-garden-furniture-i00-standard--xfwszou",
+ "taskKey": "behavior-1k/carrying_out_garden_furniture",
+ "dag_sha256": "1d06f3b5a0c891414123553d44f860000675991a757c0131c143640e6ba130b3",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 774,
+   "sim_steps": 827,
+   "physics_steps": 3308,
+   "steps": 97,
+   "elapsed_s": 6173.3,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 53,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 1,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.0,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.255,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.272,
+   "peak_cpus": 1.59,
+   "peak_pids": 150,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-carrying-out-garden-furniture-i00-standard--xfwszou.mp4",
+  "duration_s": 27.6,
+  "fps": 10.0,
+  "frames": 276,
+  "sha256": "cbf2ae104919cb77c45776b2f86ff3bb523a0277e3db4fa999abb996e8d2fc92"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "chair_on_garden_floor": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 27.0,
+      "log_step": 86,
+      "text_zh": "观察：视频 0-27.6 s 全程只见机器人在室内木地板上转动躯干、移动并在约视频 18 s 起倾倒，夹爪从未接触草坪椅，草坪椅未被搬出屋；评分器 q_score=0，success=0。"
+     }
+    ]
+   },
+   "wheelbarrow_on_lawn": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 27.0,
+      "log_step": 86,
+      "text_zh": "观察：视频 0-27.6 s 全程只见机器人在室内木地板上转动躯干、移动并在约视频 18 s 起倾倒，夹爪从未接触独轮车，独轮车未被搬出屋；评分器 q_score=0，success=0。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "比例",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "chair_on_garden_floor",
+   "where": "转场",
+   "t": 18.0,
+   "log_step": 86,
+   "evidence_zh": "观察：视频约 18 s 机器人在靠近露台门处底盘/躯干翻倒，相机画面转为侧倾/倒置，此后至 27.6 s 一直倾倒，从未抓取任何家具。log step 86 自述去露台门前测试通行，随后 step 88-205 都在尝试恢复直立。"
+  },
+  "phenomena": [
+   {
+    "t": 18.0,
+    "log_step": 88,
+    "evidence_zh": "观察：视频约 17-19 s 画面由俯视地板翻转为倾斜/倒置视角，机器人倒地后未恢复。",
+    "cat": "全身",
+    "item": "平衡"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "无状态日志；倒下的直接原因（门框碰撞或躯干指令）视频看不出，故不给根因。agent 前 1800 s 以上都在做网络探查与建图，未尝试抓取，但这不作为根因。",
+  "unmet": {
+   "chair_on_garden_floor": false,
+   "wheelbarrow_on_lawn": false
+  }
+ },
+ "summary_zh": "观察：视频 27.6 s 内机器人从未抓取草坪椅或独轮车，两个目标节点均未达成（q_score 0）。约视频 18 s 机器人在露台门附近倾倒，之后没有恢复，断点为第一个放置步骤（转场中）。无状态日志，倒下原因未能确定，根因留空。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "两个节点均未达成（q_score=0）、约视频18 s倾倒、断点与全身·平衡现象均与帧一致；根因留空并附gap_zh合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:53:15Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-carrying-out-garden-furniture-i00-standard--xfwszou/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-carrying-out-garden-furniture-i00-standard--xfwszou/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "d3c648447b61ebcadca48fd3537d3f944f600e841a971949ae81c39f48e80f45",
+   "video_sha256": "cbf2ae104919cb77c45776b2f86ff3bb523a0277e3db4fa999abb996e8d2fc92",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/clean_a_trumpet/trajectories/behavior-clean-a-trumpet-i00-standard--3ybjnmy.json b/phaseR/behavior-1k/clean_a_trumpet/trajectories/behavior-clean-a-trumpet-i00-standard--3ybjnmy.json
new file mode 100644
index 0000000000000000000000000000000000000000..dbdc35372a24d2cdadd93d8fbbf73f9c8fafad9a
--- /dev/null
+++ b/phaseR/behavior-1k/clean_a_trumpet/trajectories/behavior-clean-a-trumpet-i00-standard--3ybjnmy.json
@@ -0,0 +1,176 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-clean-a-trumpet-i00-standard--3ybjnmy",
+ "taskKey": "behavior-1k/clean_a_trumpet",
+ "dag_sha256": "2bbd17ed97252e1f070d2dc8ee7fd090c0b4c81c70149c5404c42241513c02b9",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 22591,
+   "sim_steps": 22591,
+   "physics_steps": 90364,
+   "steps": 21844,
+   "elapsed_s": 12647.4,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "q_score": 0.0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-clean-a-trumpet-i00-standard--3ybjnmy/replay.mp4",
+  "duration_s": 753.034,
+  "fps": 30.0,
+  "frames": 22591,
+  "sha256": "806ef8623ffb4e50b65018bbb220b38f0c7fa243fc8cee23760ce75bdf17bf78"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cornet_clean": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 752.0,
+      "log_step": 831,
+      "text_zh": "观察：视频 752 s 末帧腕部相机中夹爪仍夹着小号管身、另一夹爪持红色刷子，未见回合成功结束；评分器 q_score=0、success=0（grader 字段）。小号上的灰尘颗粒在头部/腕部视角无法逐粒辨认，故只依评分器总分判定未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "压/擦",
+      "note_zh": "观察：视频多处（如 60-700 s 抽样帧）一只夹爪持红色刷子在小号上方/管身、喇叭口附近擦扫。"
+     },
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频约 175-200 s 小号从书桌边缘落到地面（agent 自述 log step 225 与之吻合），随后另一夹爪把小号拾起并持握（视频 752 s 仍夹着）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：视频中底盘多次绕书桌/椅子移动，持握期间亦移动。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "cornet_clean",
+   "t": null,
+   "log_step": 831,
+   "where": "操作中",
+   "evidence_zh": "观察：全程约 753 s 视频均在用刷子清扫小号（部分时间小号被夹爪持握并移动），结束时回合未完成（q_score 0）；末步 log step 830 agent 自述正把小号前移以避免与机器人接触，仍在操作。"
+  },
+  "unmet": {
+   "cornet_clean": false
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无状态日志，灰尘颗粒在视频中不可辨，无法确定哪些区域未除尘、失败的具体现象与根因；观察：视频约 175-200 s 小号从书桌边缘滑落到地面（agent 自述 log step 225 吻合），随后被夹爪拾起，直到回合结束仍持握，而成功判据要求小号离手。"
+ },
+ "summary_zh": "观察：机器人用刷子清扫小号，中途小号滑落桌沿，之后改为单手持小号继续清扫，至视频 753 s 回合结束时仍未完成（q_score 0），唯一节点“小号除尘”未达成。断点为该节点的操作中；因无状态日志且灰尘不可辨，未给出现象和根因。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点 cornet_clean 未达成、q_score=0、断点 ref 与 where 合理，空现象/根因并附 gap_zh 对灰尘不可辨的情形可以接受。唯一需改动：gap_zh 与接触备注称小号滑落桌沿“未在视频中定位”，但视频中可看到该事件，应给出视频时间。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "唯一节点 cornet_clean 未达成（q_score=0）、断点 ref/where 合理；小号滑落桌沿并被夹爪持握的描述与视频一致；灰尘不可辨，空现象/根因并附 gap_zh 可接受。上一轮意见已采纳。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:28Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-clean-a-trumpet-i00-standard--3ybjnmy/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-clean-a-trumpet-i00-standard--3ybjnmy/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-clean-a-trumpet-i00-standard--3ybjnmy/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-clean-a-trumpet-i00-standard--3ybjnmy/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "04e8ac6d0f5ed193809c2999f3f8bc9023a2fd51872e43fec6627f2c4bcdee1c",
+   "video_sha256": "806ef8623ffb4e50b65018bbb220b38f0c7fa243fc8cee23760ce75bdf17bf78",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/clean_your_rusty_garden_tools/trajectories/behavior-clean-your-rusty-garden-tools-i00-standard--jjzb5gu.json b/phaseR/behavior-1k/clean_your_rusty_garden_tools/trajectories/behavior-clean-your-rusty-garden-tools-i00-standard--jjzb5gu.json
new file mode 100644
index 0000000000000000000000000000000000000000..291a4375e75327bd8e40c09fa995fee7b9c46fb6
--- /dev/null
+++ b/phaseR/behavior-1k/clean_your_rusty_garden_tools/trajectories/behavior-clean-your-rusty-garden-tools-i00-standard--jjzb5gu.json
@@ -0,0 +1,273 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-clean-your-rusty-garden-tools-i00-standard--jjzb5gu",
+ "taskKey": "behavior-1k/clean_your_rusty_garden_tools",
+ "dag_sha256": "af817db290d8d169ff928c865c2e9fa0b2cf6c1d8a0bfe34e98149df4f4a1029",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 2337,
+   "sim_steps": 2487,
+   "physics_steps": 9948,
+   "steps": 751,
+   "elapsed_s": 6490.8,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 150,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.0,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.334,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.299,
+   "peak_cpus": 0.99,
+   "peak_pids": 119,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-clean-your-rusty-garden-tools-i00-standard--jjzb5gu.mp4",
+  "duration_s": 83.0,
+  "fps": 10.0,
+  "frames": 830,
+  "sha256": "86f60a8b3024617367c1e9ca965ba6bb6dcad25a1fd34aac1a90a4d6f1382d5c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 6,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "toolbox_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 70.0,
+      "log_step": 314,
+      "text_zh": "观察：视频 62–82.9 s 的腕部相机里工具箱（红色）盖始终贴合箱体，没有露出内腔；右夹爪夹着箱体把手，托盘类视角未见盖打开。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：右夹爪夹住工具箱把手（视频约 66–83 s），未能掀开盖"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动到工具箱旁"
+     }
+    ]
+   },
+   "trowel_rust_removed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 70.0,
+      "log_step": 314,
+      "text_zh": "观察：视频全程未见钢丝绒擦拭园艺铲；除锈状态在腕部画面中不可判定，grader q_score=0。"
+     }
+    ]
+   },
+   "scraper_rust_removed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 70.0,
+      "log_step": 314,
+      "text_zh": "观察：视频全程未见擦拭刮刀；grader q_score=0（推断：无目标文字满足）。"
+     }
+    ]
+   },
+   "trowel_in_toolbox": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 70.0,
+      "log_step": 314,
+      "text_zh": "观察：视频 50–74 s 左夹爪一直夹着黑色铲子，未入箱；82.5 s 铲子在箱外（log step 314 仍在“放置铲子并摆正工具箱”）。"
+     }
+    ]
+   },
+   "scraper_in_toolbox": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 70.0,
+      "log_step": 314,
+      "text_zh": "观察：视频中未见刮刀入箱。"
+     }
+    ]
+   },
+   "toolbox_closed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 70.0,
+      "log_step": 314,
+      "text_zh": "观察：箱盖始终贴合，但前置节点未达成，不判定为达成。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "比例",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "toolbox_opened",
+   "t": 70.0,
+   "log_step": 314,
+   "where": "操作中",
+   "evidence_zh": "观察：约 66–83 s 右夹爪夹着工具箱把手、左夹爪仍夹着铲子，箱盖始终闭合，视频结束时仍未打开；“为双臂开盖而放铲子/调整工具箱”仅来自 agent 自述（log step 314），属推断。"
+  },
+  "unmet": {
+   "trowel_rust_removed": false,
+   "scraper_rust_removed": false,
+   "trowel_in_toolbox": false,
+   "scraper_in_toolbox": false,
+   "toolbox_closed": false
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无状态日志，视角多被箱体遮挡，无法确定盖打不开的具体现象与根因（没建立有效开盖接触 vs 位姿不准）；视频仅 83 s 而 agent 会话约 7000 s，agent 自述不作现象依据。"
+ },
+ "summary_zh": "观察：视频 83 s 内机器人抓着工具箱把手，箱盖始终未打开，未见除锈、入箱，q_score=0，0/6 节点达成。断点在打开工具箱（视频约 70 s），具体现象与根因因遮挡和无状态日志未能判定。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "0/6节点、断点 toolbox_opened、空 phenomena 加 gap_zh 基本成立；但证据文字有事实错误：视频 62–83 s 左夹爪一直夹着铲子（并非铲子在箱外地面上），右夹爪夹把手在约 70 s 才可见，62 s 时右腕相机只见地面。",
+    "n_issues": 2,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "0/6 节点、断点 toolbox_opened、空 phenomena 加 gap_zh 与视频一致；上轮证据修正已落实，未发现新错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:23Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-clean-your-rusty-garden-tools-i00-standard--jjzb5gu/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-clean-your-rusty-garden-tools-i00-standard--jjzb5gu/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-clean-your-rusty-garden-tools-i00-standard--jjzb5gu/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-clean-your-rusty-garden-tools-i00-standard--jjzb5gu/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ff37bbd4d70872b82df84afa7afa5e13d64f07824fc24b9bed5aa233312af03b",
+   "video_sha256": "86f60a8b3024617367c1e9ca965ba6bb6dcad25a1fd34aac1a90a4d6f1382d5c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/cleaning_up_plates_and_food/trajectories/behavior-cleaning-up-plates-and-food-i00-privileged--nfx7atc.json b/phaseR/behavior-1k/cleaning_up_plates_and_food/trajectories/behavior-cleaning-up-plates-and-food-i00-privileged--nfx7atc.json
new file mode 100644
index 0000000000000000000000000000000000000000..336353ecf2e0ddd769514cf3e73e57131bbf60c0
--- /dev/null
+++ b/phaseR/behavior-1k/cleaning_up_plates_and_food/trajectories/behavior-cleaning-up-plates-and-food-i00-privileged--nfx7atc.json
@@ -0,0 +1,352 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-cleaning-up-plates-and-food-i00-privileged--nfx7atc",
+ "taskKey": "behavior-1k/cleaning_up_plates_and_food",
+ "dag_sha256": "4390d52c232c8e81af0f654f0ec8aa4dc44a121bf0f5b36cb4eec4e4dab264ef",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "all_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 23241,
+   "sim_breakdown": 0,
+   "goal": 1,
+   "settle_steps": 0,
+   "goal_first_step": 23212,
+   "sim_steps": 23241,
+   "fallen": 0,
+   "final_reward": 1.0,
+   "q_score": 1.0,
+   "replay_wall_s": 4767.5,
+   "replay_timeout_s": 16694.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 2,
+   "agent_resets": 1,
+   "agent_sim_steps": 0,
+   "agent_records": 1,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.544,
+   "peak_memory_gb": 14.663,
+   "peak_processes": 48,
+   "peak_disk_gb": 2.063,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-cleaning-up-plates-and-food-i00-privileged--nfx7atc.mp4",
+  "duration_s": 774.7,
+  "fps": 30.0,
+  "frames": 23241,
+  "sha256": "fe058936711848fa9df81954c72b091e6cf6868a706891609ffdccdca8604ea9"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 8,
+  "base": 3,
+  "peak": 8,
+  "final": 8,
+  "nodes": {
+   "pizzas_on_plates_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：视频 0 s 起两份披萨各在餐桌上不同餐盘中；视频 690-760 s 机器人端着盘子把披萨送入冰箱，披萨始终在盘上，直至视频结束。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "托/舀",
+      "note_zh": "餐盘从下方承托披萨；夹爪抓握餐盘边缘搬运（观察：视频 400-700 s 腕部相机）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驱动，搬运餐盘穿过厨房"
+     }
+    ]
+   },
+   "pizzas_on_plates_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：视频 0 s 起两份披萨各在餐桌上不同餐盘中；视频 690-760 s 机器人端着盘子把披萨送入冰箱，披萨始终在盘上，直至视频结束。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "托/舀",
+      "note_zh": "餐盘从下方承托披萨；夹爪抓握餐盘边缘搬运（观察：视频 400-700 s 腕部相机）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驱动，搬运餐盘穿过厨房"
+     }
+    ]
+   },
+   "refrigerator_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      175,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 175,
+      "log_step": null,
+      "text_zh": "观察：视频约 154-175 s 机器人在冰箱前，门被打开；视频 400 s 起冰箱门已明显敞开。推断：开门时间约 175 s，未逐帧核对，不确定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "夹爪拉/推冰箱门（推断，开门瞬间未清楚看到）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在冰箱前移动"
+     }
+    ]
+   },
+   "pizzas_in_fridge_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      690,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 690,
+      "log_step": null,
+      "text_zh": "观察：视频 690 s 头部相机见冰箱内置物架上已有一份盘上披萨，机器人手中另持一盘；约 ±15 s 不确定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "托/舀",
+      "note_zh": "夹爪持盘把披萨送入冰箱"
+     }
+    ]
+   },
+   "pizzas_in_fridge_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      752,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 752,
+      "log_step": null,
+      "text_zh": "观察：视频 768-775 s 冰箱内两层各有一盘披萨，机器人在关门；第二份约 750 s 放入，约 ±10 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "托/舀",
+      "note_zh": "夹爪持盘把第二份披萨送入冰箱"
+     }
+    ]
+   },
+   "bowls_in_sink_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      66,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 66,
+      "log_step": null,
+      "text_zh": "观察：视频 46-70 s 腕部相机见白碗被带到厨房水槽上方，随后机器人带空手返回餐桌；约 ±10 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓握碗边搬到水槽"
+     }
+    ]
+   },
+   "bowls_in_sink_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      142,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 142,
+      "log_step": null,
+      "text_zh": "观察：视频 136-142 s 腕部相机见第二个碗在水槽处；约 ±10 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓握碗边搬到水槽"
+     }
+    ]
+   },
+   "refrigerator_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      175
+     ],
+     [
+      773.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 773.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0 s 冰箱关闭，175 s 起打开；视频 768-775 s 机器人手臂推门，门接近关位。推断：评分器 goal=1、q=1.0 要求冰箱关闭，故末态 true；视频最后一帧门是否完全贴合无法逐帧确认。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "夹爪推冰箱门关上"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 1.0,
+   "unit": "1",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "观察：评分器 success=1、q_score=1.0；视频显示两个碗送入水槽（约 66、142 s）、冰箱打开（约 175 s）、两盘披萨先后放入冰箱（约 690、752 s）、末段（视频 768-775 s）机器人关冰箱门，8 个节点全部 true。无状态日志，节点时间为视频粗略估计（头部相机视角常遮挡，约 ±10-30 s）；冰箱开门与关门完成的精确时刻最不确定。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "成功回合（q=1.0，goal=1），8个节点全true，起始三个节点从0开始、其余时间在视频抽帧下基本一致；无失败层，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:54:43Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-cleaning-up-plates-and-food-i00-privileged--nfx7atc/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cleaning-up-plates-and-food-i00-privileged--nfx7atc/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "bfb4e8112ba18366025ced3979297eb1a47f1d1150b3cf4fc15783aced166835",
+   "video_sha256": "fe058936711848fa9df81954c72b091e6cf6868a706891609ffdccdca8604ea9",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/cleaning_up_plates_and_food/trajectories/behavior-cleaning-up-plates-and-food-i00-privileged--uy3ezze.json b/phaseR/behavior-1k/cleaning_up_plates_and_food/trajectories/behavior-cleaning-up-plates-and-food-i00-privileged--uy3ezze.json
new file mode 100644
index 0000000000000000000000000000000000000000..8ea1eca0ca423ac3d93b3ad378a11bec9bbe859c
--- /dev/null
+++ b/phaseR/behavior-1k/cleaning_up_plates_and_food/trajectories/behavior-cleaning-up-plates-and-food-i00-privileged--uy3ezze.json
@@ -0,0 +1,320 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-cleaning-up-plates-and-food-i00-privileged--uy3ezze",
+ "taskKey": "behavior-1k/cleaning_up_plates_and_food",
+ "dag_sha256": "4390d52c232c8e81af0f654f0ec8aa4dc44a121bf0f5b36cb4eec4e4dab264ef",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 0,
+   "n_actions": 328,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 478,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 1023.4,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 0,
+   "agent_sim_steps": 478,
+   "agent_records": 0,
+   "agent_replays": 1,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 11.785,
+   "peak_memory_gb": 23.668,
+   "peak_processes": 80,
+   "peak_disk_gb": 2.211,
+   "peak_cpus": 8.1,
+   "peak_pids": 763,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-cleaning-up-plates-and-food-i00-privileged--uy3ezze.mp4",
+  "duration_s": 15.934,
+  "fps": 30.0,
+  "frames": 478,
+  "sha256": "ce2795dc8b52c30a70887cc63dc14faaca314cea06c52abee88f5d853b9fe48f"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 8,
+  "base": 3,
+  "peak": 3,
+  "final": 3,
+  "nodes": {
+   "pizzas_on_plates_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 15.9,
+      "log_step": null,
+      "text_zh": "观察：视频 0 s 与 15.9 s 帧中两份披萨都在各自盘上、盘在餐桌上，全程未见离盘。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "开局即成立，视频 0–15.9 s 未见夹爪触盘；agent 自述的抓握不采信"
+     }
+    ]
+   },
+   "pizzas_on_plates_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 15.9,
+      "log_step": null,
+      "text_zh": "观察：视频 15.9 s 末帧两盘两披萨仍在餐桌上并保持配对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "开局即成立，视频 0–15.9 s 未见夹爪触盘；agent 自述的抓握不采信"
+     }
+    ]
+   },
+   "refrigerator_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 15.9,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机）未见冰箱门打开、披萨入箱或碗入水槽；末帧碗仍在餐桌上。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频未见任何开门接触，冰箱门全程关闭；agent 自述不采信"
+     }
+    ]
+   },
+   "pizzas_in_fridge_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 15.9,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机）未见冰箱门打开、披萨入箱或碗入水槽；末帧碗仍在餐桌上。"
+     }
+    ]
+   },
+   "pizzas_in_fridge_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 15.9,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机）未见冰箱门打开、披萨入箱或碗入水槽；末帧碗仍在餐桌上。"
+     }
+    ]
+   },
+   "bowls_in_sink_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 15.9,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机）未见冰箱门打开、披萨入箱或碗入水槽；末帧碗仍在餐桌上。"
+     }
+    ]
+   },
+   "bowls_in_sink_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 15.9,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机）未见冰箱门打开、披萨入箱或碗入水槽；末帧碗仍在餐桌上。"
+     }
+    ]
+   },
+   "refrigerator_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 15.9,
+      "log_step": null,
+      "text_zh": "观察：视频中冰箱始终未打开（开局即关闭，未被操作），按状态节点单独判断成立；q_score 0 不计开局已成立条件。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "1",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "refrigerator_opened",
+   "t": null,
+   "where": "首步前",
+   "evidence_zh": "观察：视频 0–15.9 s 机器人只移动/转向，冰箱门未开，两盘未被抬起、披萨留在桌上；agent 自述的抓盘抬起未在交付视频中出现。",
+   "log_step": null
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无状态日志，头部相机下机器人未触及盘/冰箱，现象与根因证据不足故留空；agent 自述（log step 242：够不着、门太窄、披萨滑落）与交付视频不符，不采信。",
+  "unmet": {
+   "pizzas_on_plates_1": true,
+   "pizzas_on_plates_2": true,
+   "pizzas_in_fridge_1": false,
+   "pizzas_in_fridge_2": false,
+   "bowls_in_sink_1": false,
+   "bowls_in_sink_2": false,
+   "refrigerator_closed": true
+  }
+ },
+ "summary_zh": "观察：视频 15.9 s 中机器人只移动/转向，未触及盘或冰箱；两份披萨始终在盘上（开局即成立），冰箱未开、披萨未入箱、碗未入水槽，卡在第一步开冰箱门之前。q_score=0 不计开局已成立的3个节点，与之不矛盾。agent 自述（log step 242）的抬盘/披萨滑落与视频不符，现象与根因留空（证据不足）。无状态日志。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点值基本对（开局成立的3个节点、其余未达成），但 contact_actual 和断点证据把 agent 自述当成交付视频事实：视频中机器人全程只是转动/移动，两盘从未被抬起；断点应落在第一个失败步骤 refrigerator_opened 而非 pizzas_in_fridge_1。",
+    "n_issues": 4,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点值（开局成立的3个节点、其余未达成）、断点 refrigerator_opened/首步前、现象与根因留空并附 gap_zh 均与视频一致；视频中冰箱门全程关闭、盘与披萨未动、碗仍在桌上，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:29Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-cleaning-up-plates-and-food-i00-privileged--uy3ezze/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cleaning-up-plates-and-food-i00-privileged--uy3ezze/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cleaning-up-plates-and-food-i00-privileged--uy3ezze/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-cleaning-up-plates-and-food-i00-privileged--uy3ezze/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "1525b28c33e059ba4c1e49774eeb5a5fddf38d377c07572ee62daf218a542114",
+   "video_sha256": "ce2795dc8b52c30a70887cc63dc14faaca314cea06c52abee88f5d853b9fe48f",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/clearing_food_from_table_into_fridge/trajectories/behavior-clearing-food-from-table-into-fridge-i00-standard--bhwcdve.json b/phaseR/behavior-1k/clearing_food_from_table_into_fridge/trajectories/behavior-clearing-food-from-table-into-fridge-i00-standard--bhwcdve.json
new file mode 100644
index 0000000000000000000000000000000000000000..11b0c562c229db1a00e45538cda89b1ca063ab0b
--- /dev/null
+++ b/phaseR/behavior-1k/clearing_food_from_table_into_fridge/trajectories/behavior-clearing-food-from-table-into-fridge-i00-standard--bhwcdve.json
@@ -0,0 +1,305 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-clearing-food-from-table-into-fridge-i00-standard--bhwcdve",
+ "taskKey": "behavior-1k/clearing_food_from_table_into_fridge",
+ "dag_sha256": "0187c62673ba112bb7fbae8d19e1bccdbef83deaf34d00b39cbfc9668575af71",
+ "outcome": {
+  "success": 0,
+  "reward": 0.4,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 25574,
+   "sim_steps": 25574,
+   "physics_steps": 102296,
+   "steps": 25544,
+   "elapsed_s": 13705.1,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "q_score": 0.4,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-clearing-food-from-table-into-fridge-i00-standard--bhwcdve/replay.mp4",
+  "duration_s": 852.467,
+  "fps": 30.0,
+  "frames": 25574,
+  "sha256": "27e5e3815b12bb0666465cd2903a660ae850b2067420000192d142234d5bbf92"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 6,
+  "base": 1,
+  "peak": 3,
+  "final": 3,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      352,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 345,
+      "log_step": null,
+      "text_zh": "观察：视频 345 s 机器人手爪抓住冰箱门把手；355 s 门板已离开柜体、内腔露出，之后一直开着（视频 525 s、850 s 仍开）。时间约 ±10 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "机器人夹爪抓住冰箱把手拉门"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在冰箱前调整站位"
+     }
+    ]
+   },
+   "chicken_in_tupperware": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 190,
+      "log_step": null,
+      "text_zh": "观察：视频约 160-230 s 机器人松手后鸡肉落在台面保鲜盒旁；之后鸡肉掉到地上（视频约 260 s），多次抓取失败；视频 800-852 s 腕部相机里鸡肉仍夹在夹爪中、不在盒内。推断：全程未见鸡肉入盒；评分 q=0.4 也只容许 2 个目标成立（派、盒1），与此一致。日志 step 848 自述鸡肉已入盒，与视频不符，不采信。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪多次抓握鸡肉，多次松手后落在盒外/地上"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动并支承"
+     }
+    ]
+   },
+   "pie_in_tupperware": {
+    "method": "visual",
+    "intervals": [
+     [
+      750,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 750,
+      "log_step": 789,
+      "text_zh": "观察：视频约 745-765 s 左手夹爪将苹果派放入右手持的保鲜盒内，765 s 画面可见盒内有派（约 ±10 s）；随后盒被提起、放地，派仍在盒内。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "左夹爪抓握派放入盒内（先前多次滑脱）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动靠近桌子"
+     }
+    ]
+   },
+   "tupperwares_in_fridge_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      520,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 525,
+      "log_step": 524,
+      "text_zh": "观察：视频 525 s 头部相机可见一个白色保鲜盒放在冰箱内搁板上；更早放入时曾滑落到地上（agent 自述，log step 413）。推断：盒1自约 520 s 起在冰箱内，到结束未取出；q=0.4 与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓握盒沿，多次放入冰箱，首次滑落后重放"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在冰箱前调整位置"
+     }
+    ]
+   },
+   "tupperwares_in_fridge_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 852,
+      "log_step": 899,
+      "text_zh": "观察：视频末段（800-852 s）机器人在冰箱前，盒2由手持着，未放入冰箱；视频到结束时只有盒1在冰箱内。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪抓握盒沿，搬运到冰箱前并尝试推入搁板"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘搬运途中移动"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      352
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：视频开头冰箱门关闭；352 s 起打开，之后到视频结束（852 s）门始终敞开，未关回。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.4,
+   "unit": "ratio",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "chicken_in_tupperware",
+   "t": 190,
+   "log_step": 126,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 160-260 s 鸡肉松手后落在盒旁并掉到地上，之后在地面/台面反复抓取，直到视频结束始终没入盒（视频 800-852 s 鸡肉仍在夹爪中）；机器人转去处理盒子和派，这一步从未补上。agent 自述“鸡肉落在盒旁”（log step 126）。"
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 190,
+    "log_step": 126,
+    "evidence_zh": "观察：视频约 190 s 松手时鸡肉在台面盒旁，不在盒内（agent 自述 log step 126、139 同）；推断：松手前未到位，按 D24 记位姿不准。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "alt_feasible": "未知",
+    "evidence_zh": "观察：盒子轻、易滑动倾倒，agent 自述盒滑动、翻倒、从台面端掉落（log step 142、183、305），鸡肉形状不规则、窄端抓取多次滑脱（log 253-268、837）；推断：放置与抓取方案不稳，子类不明，停在控制。",
+    "source_zh": "视频 / agent 自述"
+   }
+  ],
+  "gap_zh": "无状态日志，头部相机常看不到盒内；盒1入冰箱的起始时间、派入盒的时间约 ±10 s；鸡肉在 800 s 后是否入盒按腕部相机判断为否，agent 自述相反。",
+  "unmet": {
+   "chicken_in_tupperware": false,
+   "pie_in_tupperware": true,
+   "tupperwares_in_fridge_1": true,
+   "tupperwares_in_fridge_2": false,
+   "fridge_closed": false
+  }
+ },
+ "summary_zh": "观察：视频 852 s 内机器人约 352 s 打开冰箱，约 520 s 把一个保鲜盒放进冰箱，约 750 s 把派装入另一盒，但鸡肉始终没入盒，盒2到结束未进冰箱，冰箱未关（q=0.4，与派+盒1两个目标一致）。断点为鸡肉入盒：松手后鸡肉落在盒旁/地上，反复抓取失败；根因仅能到控制层。缺口：无状态日志，部分时间为估计。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 5,
+    "summary_zh": "节点值（开冰箱、派入盒、盒1入冰箱为真；鸡肉入盒、盒2入冰箱、关冰箱为假）与视频抽帧及 q=0.4（5个目标中2个）一致；断点取最早未修复的鸡肉入盒合理，未发现需要改动的错误。根因“控制”证据主要来自 agent 自述，较弱，但标注为推断且 conf 偏低，可接受。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:55:05Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-clearing-food-from-table-into-fridge-i00-standard--bhwcdve/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-clearing-food-from-table-into-fridge-i00-standard--bhwcdve/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "d373eb675771b425722642125a94a6724489714ab68998c6b5b158b015f0558e",
+   "video_sha256": "27e5e3815b12bb0666465cd2903a660ae850b2067420000192d142234d5bbf92",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/clearing_food_from_table_into_fridge/trajectories/behavior-clearing-food-from-table-into-fridge-i00-standard--hr5bciy.json b/phaseR/behavior-1k/clearing_food_from_table_into_fridge/trajectories/behavior-clearing-food-from-table-into-fridge-i00-standard--hr5bciy.json
new file mode 100644
index 0000000000000000000000000000000000000000..cff93fe413a152c8712397fdc4235b0faaf6f4dd
--- /dev/null
+++ b/phaseR/behavior-1k/clearing_food_from_table_into_fridge/trajectories/behavior-clearing-food-from-table-into-fridge-i00-standard--hr5bciy.json
@@ -0,0 +1,359 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-clearing-food-from-table-into-fridge-i00-standard--hr5bciy",
+ "taskKey": "behavior-1k/clearing_food_from_table_into_fridge",
+ "dag_sha256": "0187c62673ba112bb7fbae8d19e1bccdbef83deaf34d00b39cbfc9668575af71",
+ "outcome": {
+  "success": 0,
+  "reward": 0.4,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 6006,
+   "sim_steps": 6156,
+   "physics_steps": 24624,
+   "steps": 2002,
+   "elapsed_s": 6779.5,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 150,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.4,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "peak_cpus": 0.0,
+   "peak_pids": 0.0,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-clearing-food-from-table-into-fridge-i00-standard--hr5bciy.mp4",
+  "duration_s": 205.3,
+  "fps": 10.0,
+  "frames": 2053,
+  "sha256": "dd4317562d69b403cf90e7a1f085b635d39fa4050370789b5a96fd4f35d5f551"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 6,
+  "base": 1,
+  "peak": 4,
+  "final": 3,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      115.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 120.0,
+      "log_step": null,
+      "text_zh": "观察：视频 120 s 头部相机可见冰箱门板离开柜体、内腔露出；视频 100–110 s 机器人仍端着盒子在桌边/途中，门开启时刻约 115 s（不确定 ±5 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：夹爪抓握冰箱门把手拉开门板（门开启瞬间被遮挡，推断）；底盘移动站位"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘在搬运过程中移动并支承机器人"
+     }
+    ]
+   },
+   "chicken_in_tupperware": {
+    "method": "visual",
+    "intervals": [
+     [
+      105.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 110.0,
+      "log_step": null,
+      "text_zh": "观察：视频 98 s 夹爪从盘上取鸡肉，视频 110 s、120 s 头部相机可见盒内有烤鸡并被端往冰箱，至视频 190 s 冰箱内盒中仍有鸡肉。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：一只夹爪抓握鸡肉放入另一只夹爪端着的保鲜盒"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘在搬运过程中移动并支承机器人"
+     }
+    ]
+   },
+   "pie_in_tupperware": {
+    "method": "visual",
+    "intervals": [
+     [
+      160.0,
+      199.0
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 168.0,
+      "log_step": null,
+      "text_zh": "观察：视频 84 s 派在盘上；视频 168 s、192 s、198 s 夹爪端着第二个盒，盒内有白色派状物，桌上盘子已空。推断：派已入盒，约 160 s（不确定）。派入盒仅持续到盒掉落（约 199 s）。"
+     },
+     {
+      "t": 199.0,
+      "log_step": null,
+      "text_zh": "观察：视频 199 s 腕部相机可见倒地的盒子，盒外地面上有黄色派片，派已洒出盒外；201–205 s 仍在盒外。q_score=0.4 对应 5 个字面中 2 个成立（鸡肉入盒、盒1入冰箱），与派终态不成立一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：夹爪抓握苹果派放入第二个保鲜盒（视频 84–160 s 间，具体放入时刻被遮挡）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘在搬运过程中移动并支承机器人"
+     }
+    ]
+   },
+   "tupperwares_in_fridge_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      148.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 190.0,
+      "log_step": null,
+      "text_zh": "观察：视频 150 s 起装鸡肉的盒放在冰箱上层隔板，视频 186 s、190 s、205 s 头部相机仍在冰箱内。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：夹爪抓握盒子送入冰箱隔板；底盘驶近冰箱"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘在搬运过程中移动并支承机器人"
+     }
+    ]
+   },
+   "tupperwares_in_fridge_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 204.0,
+      "log_step": null,
+      "text_zh": "观察：视频 198 s 起第二个盒被端到冰箱旁，视频 204 s 头部相机可见白色盒落在冰箱前地面，冰箱内只有一个盒。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：夹爪抓握第二个盒（含派）运往冰箱，中途掉到地上"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘在搬运过程中移动并支承机器人"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      115.0
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 205.0,
+      "log_step": null,
+      "text_zh": "观察：冰箱初始关闭；视频 120 s 起门敞开，直到视频 205 s 结束仍敞开，未见关门。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.4,
+   "unit": "ratio",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "tupperwares_in_fridge_2",
+   "t": 198.0,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 168–198 s 机器人端着含派的第二个盒走向冰箱；视频 198–204 s 该盒落在冰箱前地面，没有进入冰箱，之后到视频 205 s 结束都没有重新拿起，冰箱门也一直敞开。观察：视频 199 s 起盒外地面可见派片，派洒出盒外。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 198.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 192 s 夹爪仍端着第二个盒，视频 198 s 盒贴近冰箱门底沿，视频 204 s 盒落在地上。推断：搬运途中盒子脱手（非计划松手）。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "视频只有 205 s，且手腕相机常被遮挡，掉盒的具体原因（夹持过松、被冰箱门挡到）看不清；无状态日志；log 是 7000 s 的会话，不能对应视频，故不给根因。",
+  "unmet": {
+   "chicken_in_tupperware": true,
+   "pie_in_tupperware": false,
+   "tupperwares_in_fridge_1": true,
+   "tupperwares_in_fridge_2": false,
+   "fridge_closed": false
+  }
+ },
+ "summary_zh": "观察：视频 205 s 内机器人把鸡肉装入盒并放进冰箱（约 148 s），第二个盒装了派但在运往冰箱途中掉在地上（约 198–204 s），派洒到盒外地面，冰箱始终敞开。峰值到达 打开冰箱、鸡肉入盒、派入盒、盒入冰箱(1)，终态派入盒不再成立（final 3，与 q_score 0.4=2/5 一致），断点为第二个盒入冰箱（没保持住）。无状态日志，节点时间由视频判断，根因无证据留空。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "主线（鸡肉入盒、盒1入冰箱、第二盒掉在冰箱前、断点 tupperwares_in_fridge_2 没保持住）与画面一致；但盒掉落后派片已洒到盒外地面，pie_in_tupperware 终态应为 false，final 应为 3，与 q_score=0.4（2/5 字面）也吻合。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "第1轮修改后标签与画面和 q_score=0.4 一致：峰值4、终态3，断点为第二个盒入冰箱（没保持住），根因因无证据留空并给出 gap_zh。未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T21:39:31Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work_rerun1/runs/behavior-clearing-food-from-table-into-fridge-i00-standard--hr5bciy/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work_rerun1/runs/behavior-clearing-food-from-table-into-fridge-i00-standard--hr5bciy/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work_rerun1/runs/behavior-clearing-food-from-table-into-fridge-i00-standard--hr5bciy/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work_rerun1/runs/behavior-clearing-food-from-table-into-fridge-i00-standard--hr5bciy/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "2e108ac929ef5ee42cd050f5b40b5b90f43f258d7932d71e834f2378b27ca7bf",
+   "video_sha256": "dd4317562d69b403cf90e7a1f085b635d39fa4050370789b5a96fd4f35d5f551",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/collecting_aluminum_cans/trajectories/behavior-collecting-aluminum-cans-i00-standard--ywmw6nf.json b/phaseR/behavior-1k/collecting_aluminum_cans/trajectories/behavior-collecting-aluminum-cans-i00-standard--ywmw6nf.json
new file mode 100644
index 0000000000000000000000000000000000000000..3033d58a847a03f4ee119bd09ccd57f9522ca8fb
--- /dev/null
+++ b/phaseR/behavior-1k/collecting_aluminum_cans/trajectories/behavior-collecting-aluminum-cans-i00-standard--ywmw6nf.json
@@ -0,0 +1,340 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-collecting-aluminum-cans-i00-standard--ywmw6nf",
+ "taskKey": "behavior-1k/collecting_aluminum_cans",
+ "dag_sha256": "d65ba1638f1a51ef401ef21980d53b456855008912d4b14298be41ef3e677678",
+ "outcome": {
+  "success": 0,
+  "reward": 0.8333333333333334,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 16899,
+   "sim_steps": 17028,
+   "physics_steps": 68112,
+   "steps": 5633,
+   "elapsed_s": 6582.2,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 129,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 1,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.8333333333333334,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 4.454,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.572,
+   "peak_cpus": 1.64,
+   "peak_pids": 211,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-collecting-aluminum-cans-i00-standard--ywmw6nf.mp4",
+  "duration_s": 567.7,
+  "fps": 10.0,
+  "frames": 5677,
+  "sha256": "053539859abcd518448170866f155adef10250577b38c42e3719ad18c30e413d"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 6,
+  "base": 0,
+  "peak": 5,
+  "final": 5,
+  "nodes": {
+   "cans_in_bucket_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      248.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 248.5,
+      "log_step": null,
+      "text_zh": "观察：视频约 248.5 s，机器人夹爪携蓝罐（地面罐2）到直立桶口上方，松手后桶口俯视画面（腕部/头部相机）桶内出现该罐，累计 1 罐在桶内；时间精度约 ±2 s。推断：桶壁遮挡，按放入前后连续画面计数；与评分器 q_score=5/6 一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：单侧夹爪夹住蓝罐（地面罐2）并送入桶内后张开。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘驶近桶/罐，携罐行驶期间底盘移动。"
+     }
+    ]
+   },
+   "cans_in_bucket_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      305.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 305.5,
+      "log_step": null,
+      "text_zh": "观察：视频约 305.5 s，机器人夹爪携金罐（地面罐1）到直立桶口上方，松手后桶口俯视画面（腕部/头部相机）桶内出现该罐，累计 2 罐在桶内；时间精度约 ±2 s。推断：桶壁遮挡，按放入前后连续画面计数；与评分器 q_score=5/6 一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：单侧夹爪夹住金罐（地面罐1）并送入桶内后张开。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘驶近桶/罐，携罐行驶期间底盘移动。"
+     }
+    ]
+   },
+   "cans_in_bucket_3": {
+    "method": "visual",
+    "intervals": [
+     [
+      399.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 399.5,
+      "log_step": null,
+      "text_zh": "观察：视频约 399.5 s，机器人夹爪携橙罐（床罐B1）到直立桶口上方，松手后桶口俯视画面（腕部/头部相机）桶内出现该罐，累计 3 罐在桶内；时间精度约 ±2 s。推断：桶壁遮挡，按放入前后连续画面计数；与评分器 q_score=5/6 一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：单侧夹爪夹住橙罐（床罐B1）并送入桶内后张开。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘驶近桶/罐，携罐行驶期间底盘移动。"
+     }
+    ]
+   },
+   "cans_in_bucket_4": {
+    "method": "visual",
+    "intervals": [
+     [
+      457,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 457,
+      "log_step": null,
+      "text_zh": "观察：视频约 457 s，机器人夹爪携红罐（床罐B2）到直立桶口上方，松手后桶口俯视画面（腕部/头部相机）桶内出现该罐，累计 4 罐在桶内；时间精度约 ±2 s。推断：桶壁遮挡，按放入前后连续画面计数；与评分器 q_score=5/6 一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：单侧夹爪夹住红罐（床罐B2）并送入桶内后张开。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘驶近桶/罐，携罐行驶期间底盘移动。"
+     }
+    ]
+   },
+   "cans_in_bucket_5": {
+    "method": "visual",
+    "intervals": [
+     [
+      515,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 515,
+      "log_step": null,
+      "text_zh": "观察：视频约 515 s，机器人夹爪携银/灰罐（床罐B0）到直立桶口上方，松手后桶口俯视画面（腕部/头部相机）桶内出现该罐，累计 5 罐在桶内；时间精度约 ±2 s。推断：桶壁遮挡，按放入前后连续画面计数；与评分器 q_score=5/6 一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：单侧夹爪夹住银/灰罐（床罐B0）并送入桶内后张开。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：轮式底盘驶近桶/罐，携罐行驶期间底盘移动。"
+     }
+    ]
+   },
+   "cans_in_bucket_6": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 515.5,
+      "log_step": 168,
+      "text_zh": "观察：视频 515 s 起头部画面桶内仅 5 个罐；其后机器人转向并倾倒，至 567.7 s 视频结束桶内未见第6罐（地面罐3在西南角，未被取到）。推断：评分器 q_score=5/6 也只允许 5 罐。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：视频 520-540 s 底盘驶向第6罐途中机器人倾倒；夹爪未见夹持第6罐。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.8333333333333334,
+   "unit": "比例",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cans_in_bucket_6",
+   "t": 535,
+   "log_step": 170,
+   "where": "转场",
+   "evidence_zh": "观察：视频约 520 s 第5罐入桶后，机器人驶离桶去取第6罐；视频约 535-560 s 头部相机画面倾斜并转向天空/窗户，之后到 567.7 s 结束再未恢复。log step 170 agent 自述 tilt guard 报告 up_z=0.273（严重倾倒）。第6罐未入桶。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 540,
+    "log_step": 170,
+    "evidence_zh": "观察：视频 ≈535 s 起头部相机画面逐渐倾斜，556 s 之后只见天空和窗框，直到视频结束（567.7 s）；推断：机器人在去第6罐途中倾倒。log step 170 的 up_z=0.273 为 agent 引用的 tilt guard 读数，与视频一致。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "根因证据不足：无状态日志，视频未显示倾倒时的底盘/手臂具体动作（摄像头只有头/腕视角），未标根因；各罐入桶时间由桶口俯视画面估计（±2 s）。",
+  "unmet": {
+   "cans_in_bucket_1": true,
+   "cans_in_bucket_2": true,
+   "cans_in_bucket_3": true,
+   "cans_in_bucket_4": true,
+   "cans_in_bucket_5": true,
+   "cans_in_bucket_6": false
+  }
+ },
+ "summary_zh": "观察：graded 视频 567.7 s 中机器人依次把 5 个罐放入桶（视频约 248.5、305.5、399.5、457、515 s），与 q_score=5/6 一致。第6罐（地面西南角）未取到，视频约 535 s 起机器人倾倒并持续到视频结束，断点为第6罐入桶（转场，全身·平衡）。无状态日志，入桶时间为视觉估计，根因未标。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "5 罐入桶、断点在第6罐（转场·全身平衡）与 q_score=5/6、视频倾倒（535 s 起头部画面倾斜，556 s 后只见天空）及 log step 170 的 SAFETY 读数（up_z=0.273、base_z=0.286）一致；根因留空并有 gap_zh，证据不足以支持根因，未发现需要修改的内容。各罐时间未逐一复核，仅以 q_score 约束。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:54:01Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-collecting-aluminum-cans-i00-standard--ywmw6nf/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-collecting-aluminum-cans-i00-standard--ywmw6nf/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3927642cde2fcd64bde0a17c5edec634a2e400aab2fb434fd93b62d732d8b8b7",
+   "video_sha256": "053539859abcd518448170866f155adef10250577b38c42e3719ad18c30e413d",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/collecting_childrens_toys/trajectories/behavior-collecting-childrens-toys-i00-privileged--n3emr8w.json b/phaseR/behavior-1k/collecting_childrens_toys/trajectories/behavior-collecting-childrens-toys-i00-privileged--n3emr8w.json
new file mode 100644
index 0000000000000000000000000000000000000000..fb5d9d20af49d15f5cb2074bd6f2724becec1045
--- /dev/null
+++ b/phaseR/behavior-1k/collecting_childrens_toys/trajectories/behavior-collecting-childrens-toys-i00-privileged--n3emr8w.json
@@ -0,0 +1,369 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-collecting-childrens-toys-i00-privileged--n3emr8w",
+ "taskKey": "behavior-1k/collecting_childrens_toys",
+ "dag_sha256": "c236dc50d1c43f8f6eaf6d7c859842df16d2f5e4870176f1f7cb433b6ed11266",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "all_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 11370,
+   "sim_breakdown": 0,
+   "goal": 1,
+   "settle_steps": 0,
+   "goal_first_step": 11241,
+   "sim_steps": 11370,
+   "fallen": 0,
+   "final_reward": 1.0,
+   "q_score": 1.0,
+   "replay_wall_s": 2257.8,
+   "replay_timeout_s": 8780.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 3,
+   "agent_resets": 0,
+   "agent_sim_steps": 34110,
+   "agent_records": 1,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "peak_cpus": 0.0,
+   "peak_pids": 0.0,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-collecting-childrens-toys-i00-privileged--n3emr8w.mp4",
+  "duration_s": 379.0,
+  "fps": 30.0,
+  "frames": 11370,
+  "sha256": "098239bbff3fa4afbc551d9845f61796af5d17c4dd8fb616ec3701accff08123"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 7,
+  "base": 0,
+  "peak": 7,
+  "final": 7,
+  "nodes": {
+   "dice_in_bookcase_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      97,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 97,
+      "log_step": null,
+      "text_zh": "观察（视频时钟，约 1.2 s 采样）：约 33.7 s 黑骰子第一次离开夹爪后书柜仍空、1:16 骰子在地板、1:24-1:34 重新手持；约 95 s 手空，约 99.8 s 帧书柜下层可见黑骰子；推断：入柜释放约 97 s（±3 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：机器人夹爪抓取玩具并送入书柜（粗采样帧可见手持玩具靠近书柜）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "推断：底盘在儿童房内移动于床/书桌与书柜之间，搬运时驱动。"
+     }
+    ]
+   },
+   "dice_in_bookcase_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      124,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 124,
+      "log_step": null,
+      "text_zh": "观察：白骰子手持至约 123.5 s，约 124 s 离开夹爪，128-129 s 头部相机书柜内可见两枚骰子；推断：入柜约 124 s（±2 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：机器人夹爪抓取玩具并送入书柜（粗采样帧可见手持玩具靠近书柜）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "推断：底盘在儿童房内移动于床/书桌与书柜之间，搬运时驱动。"
+     }
+    ]
+   },
+   "teddies_in_bookcase_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      161,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 161,
+      "log_step": null,
+      "text_zh": "观察：第一只泰迪熊手持至约 159 s（头部相机见手伸向书柜），约 162 s 手腕相机已空，165 s 起书柜隔板上可见泰迪熊；推断：入柜约 161 s（±3 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：机器人夹爪抓取玩具并送入书柜（粗采样帧可见手持玩具靠近书柜）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "推断：底盘在儿童房内移动于床/书桌与书柜之间，搬运时驱动。"
+     }
+    ]
+   },
+   "teddies_in_bookcase_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      201,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 201,
+      "log_step": null,
+      "text_zh": "观察：第二只泰迪熊约 197-200 s 仍在手腕相机内并靠近书柜，约 203 s 手腕相机已空；推断：入柜约 201 s（±3 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：机器人夹爪抓取玩具并送入书柜（粗采样帧可见手持玩具靠近书柜）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "推断：底盘在儿童房内移动于床/书桌与书柜之间，搬运时驱动。"
+     }
+    ]
+   },
+   "board_games_in_bookcase_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      234,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 234,
+      "log_step": null,
+      "text_zh": "观察：第一盒桌游手持至约 232 s，约 235 s 手腕相机已空且书柜隔板可见盒子；推断：入柜约 234 s（±3 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：机器人夹爪抓取玩具并送入书柜（粗采样帧可见手持玩具靠近书柜）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "推断：底盘在儿童房内移动于床/书桌与书柜之间，搬运时驱动。"
+     }
+    ]
+   },
+   "board_games_in_bookcase_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      317,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 317,
+      "log_step": null,
+      "text_zh": "观察：第二盒桌游手持至约 316.5 s，约 317.5-319 s 手腕相机已空且书柜隔板可见盒子；推断：入柜约 317 s（±2 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：机器人夹爪抓取玩具并送入书柜（粗采样帧可见手持玩具靠近书柜）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "推断：底盘在儿童房内移动于床/书桌与书柜之间，搬运时驱动。"
+     }
+    ]
+   },
+   "train_in_bookcase": {
+    "method": "visual",
+    "intervals": [
+     [
+      374.7,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 374.7,
+      "log_step": null,
+      "text_zh": "观察：火车手持靠近书柜至约 375-376 s；推断：入柜达成时刻取评分器 goal_first_step 11241 → 374.7 s（同一视频时钟，帧上仅作 ±2 s 一致性核对）；其余六件玩具此前已入柜，final 为真（goal=1、q_score=1.0）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：机器人夹爪抓取玩具并送入书柜（粗采样帧可见手持玩具靠近书柜）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "推断：底盘在儿童房内移动于床/书桌与书柜之间，搬运时驱动。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 1.0,
+   "unit": "1",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：评分器 success=1、q_score=1.0，七件玩具（骰子、泰迪熊、桌游、火车）依次搬入同一书柜，最后一件于 goal_first_step 11241（视频约 374.7 s）达成联合目标。各节点入柜时刻由约 1-3 s 采样的视频帧重估（±2-3 s，推断），无状态日志；黑骰子第一次携带（约 34 s）后并未留在书柜，约 97 s 重新放入。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "任务成功（q_score=1.0）与 final 全真无误，但六个中间节点的入柜时间明显错误，偏差远超自述的 ±15 s：视频中这些时刻玩具仍在夹爪里或尚未搬运。骰子1实际约 30 s、骰子2约 122 s、泰迪熊约 160 s 与 200 s、桌游约 230 s 与 310 s（均为视频时钟，粗估）。",
+    "n_issues": 4,
+    "n_major": 3
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "成功运行（q_score=1.0，goal_first_step 11241 对应 374.7 s）；七个节点 final 为真，抽查 100 s 帧可见黑骰子已在书柜下层，30-36 s 书柜仍空，与标注的“首次携带后未留柜、约 97 s 重新放入”一致。未发现需要修改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:19Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-collecting-childrens-toys-i00-privileged--n3emr8w/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-collecting-childrens-toys-i00-privileged--n3emr8w/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-collecting-childrens-toys-i00-privileged--n3emr8w/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-collecting-childrens-toys-i00-privileged--n3emr8w/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f2df1b35a5212ed2638d9368456915bb67685e529dc04e7755b27c4e4c3467e5",
+   "video_sha256": "098239bbff3fa4afbc551d9845f61796af5d17c4dd8fb616ec3701accff08123",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/cook_a_frozen_pie/trajectories/behavior-cook-a-frozen-pie-i00-privileged--wjogiey.json b/phaseR/behavior-1k/cook_a_frozen_pie/trajectories/behavior-cook-a-frozen-pie-i00-privileged--wjogiey.json
new file mode 100644
index 0000000000000000000000000000000000000000..928c530066fd49d1c183e44aecd2f08e1261915d
--- /dev/null
+++ b/phaseR/behavior-1k/cook_a_frozen_pie/trajectories/behavior-cook-a-frozen-pie-i00-privileged--wjogiey.json
@@ -0,0 +1,246 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-cook-a-frozen-pie-i00-privileged--wjogiey",
+ "taskKey": "behavior-1k/cook_a_frozen_pie",
+ "dag_sha256": "d00050bfce362a5142c45adf0c68a6b7072f4aa5bda1d9b540421c39c788325f",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 567,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 717,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 654.8,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 4,
+   "agent_resets": 25,
+   "agent_sim_steps": 1284,
+   "agent_records": 2,
+   "agent_replays": 1,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 17.1,
+   "peak_memory_gb": 22.187,
+   "peak_processes": 80,
+   "peak_disk_gb": 1.764,
+   "peak_cpus": 8.04,
+   "peak_pids": 642,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-cook-a-frozen-pie-i00-privileged--wjogiey.mp4",
+  "duration_s": 23.9,
+  "fps": 30.0,
+  "frames": 717,
+  "sha256": "8ddb48d13efbc780c9b1348618cb1e1968ca8fe58f1e23c52e3fc4730e6123c6"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 4,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 22.0,
+      "log_step": 233,
+      "text_zh": "观察：视频 0-23.9 s 为机器人头/腕部相机，全程在走廊/厨房驱动，视频 21-23.9 s 画面中冰箱门（灰色边框玻璃门）仍贴着箱体，未见门向外转开；agent 自述门够不着、关节角 0（log step 233）仅作旁证。"
+     }
+    ]
+   },
+   "heat_source_on": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 23.9,
+      "log_step": 233,
+      "text_zh": "观察：视频全程未见夹爪触碰任一热源控制点，画面中控制标记未变绿；评分 goal=0、q_score=0。"
+     }
+    ]
+   },
+   "pie_cooked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 23.9,
+      "log_step": 233,
+      "text_zh": "观察：派在冰箱内未取出，视频未见任何热源加热派；评分器 goal=0。"
+     }
+    ]
+   },
+   "pie_on_tray": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 23.9,
+      "log_step": 233,
+      "text_zh": "观察：视频未见派离开冰箱或落到托盘；评分器 goal=0、q_score=0。"
+     }
+    ]
+   }
+  },
+  "constraints": {},
+  "contact_actual": {
+   "fridge_opened": [
+    {
+     "who": "底盘",
+     "how": "支撑",
+     "note_zh": "观察：底盘驱动至冰箱前方走廊（视频 0-23.9 s）"
+    },
+    {
+     "who": "未知",
+     "how": "未知",
+     "note_zh": "视频画面未能显示夹爪是否触到门面；log step 177-197 自述曾按压门面但未抓住，仅为 session 证据"
+    }
+   ]
+  }
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "ratio",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "fridge_opened",
+   "t": 22.0,
+   "log_step": 233,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0-23.9 s 机器人驱动到冰箱前，冰箱门始终关闭，之后无任何后续步骤；评分 goal=0、q_score=0。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "观察：冰箱门始终未开。视频为头/腕部相机，夹爪与门的接触看不清，无状态日志；agent 自述（IK 超范围、辅助抓取失败）不采信，且 oracle 成功打开同一扇门，故现象与根因均无法由视频/评分判定。",
+  "unmet": {
+   "pie_cooked": false,
+   "pie_on_tray": false
+  }
+ },
+ "summary_zh": "观察：视频 23.9 s 内机器人只是驱动到冰箱前，冰箱门始终未打开，4 个节点均未达成（q_score 0）。断点在打开冰箱。现象与根因无法由视频/评分判定（见 gap_zh）。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点与断点（fridge_opened 未达成、4 节点 final=false）与视频和评分一致；问题在根因：控制/控制 仅来自 agent 自述（门缘超 IK 范围、辅助抓取无法触发），无视频、状态日志或评分字段支持，且 oracle 成功打开了同一冰箱门，说明并非不可达；应删除该根因并保留 gap_zh。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "4 个节点均未达成、断点在 fridge_opened 与视频和评分(goal=0,q=0)一致；现象与根因置空并有 gap_zh，符合规则。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:14Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-cook-a-frozen-pie-i00-privileged--wjogiey/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-a-frozen-pie-i00-privileged--wjogiey/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-a-frozen-pie-i00-privileged--wjogiey/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-cook-a-frozen-pie-i00-privileged--wjogiey/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3a125bdf47313a480c0a6365475e11e30db62a906fa0d33e9d1e82f5dff7e334",
+   "video_sha256": "8ddb48d13efbc780c9b1348618cb1e1968ca8fe58f1e23c52e3fc4730e6123c6",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ },
+ "conf_reason_zh": "现象与根因因证据不足留空，仅断点可由视频与评分确认。"
+}
diff --git a/phaseR/behavior-1k/cook_a_frozen_pie/trajectories/behavior-cook-a-frozen-pie-i00-standard--ntsulhv.json b/phaseR/behavior-1k/cook_a_frozen_pie/trajectories/behavior-cook-a-frozen-pie-i00-standard--ntsulhv.json
new file mode 100644
index 0000000000000000000000000000000000000000..3b6f5be635fa0ebb7843d2cefe72a47758f8e221
--- /dev/null
+++ b/phaseR/behavior-1k/cook_a_frozen_pie/trajectories/behavior-cook-a-frozen-pie-i00-standard--ntsulhv.json
@@ -0,0 +1,267 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-cook-a-frozen-pie-i00-standard--ntsulhv",
+ "taskKey": "behavior-1k/cook_a_frozen_pie",
+ "dag_sha256": "d00050bfce362a5142c45adf0c68a6b7072f4aa5bda1d9b540421c39c788325f",
+ "outcome": {
+  "success": 0,
+  "reward": 0.5,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 32817,
+   "sim_steps": 32817,
+   "physics_steps": 131268,
+   "steps": 25970,
+   "elapsed_s": 11723.8,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "q_score": 0.5,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-cook-a-frozen-pie-i00-standard--ntsulhv/replay.mp4",
+  "duration_s": 1093.9,
+  "fps": 30.0,
+  "frames": 32817,
+  "sha256": "8ecff6936c988eb5e3ab1b1ea1d1a3fde8967b175027f458a37133cf196ebdc4"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 4,
+  "base": null,
+  "peak": null,
+  "final": 3,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      44,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 44,
+      "log_step": 95,
+      "text_zh": "观察：视频 30-44 s 头部/腕部画面中冰箱右门已向外转开，露出内部玻璃层架；agent 自述“门已打开”（log step 95，control_steps 1320/30=44 s）。随后门多次回弹，视频 195 s 门大开，派可见。机器人夹爪抓握门边拉开。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：夹爪夹住冰箱门边/把手外拉"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驱动到冰箱前配合开门"
+     }
+    ]
+   },
+   "heat_source_on": {
+    "method": "visual",
+    "intervals": [
+     [
+      258,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 258,
+      "log_step": 258,
+      "text_zh": "观察：视频 255-262 s 左夹爪按烤箱控制点，指示灯由红变绿（腕部画面见绿色标记），之后全程可见绿灯。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：左夹爪指端按烤箱控制点"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驱动到烤箱前"
+     }
+    ]
+   },
+   "pie_cooked": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 1093,
+      "log_step": 839,
+      "text_zh": "推断：评分器 q_score=0.5、success=0；视频末帧派仍在托盘上且未被抓握（派在托盘上这一文字量在 q 中占一项），故“已烹熟”这一项未满足。观察：派先被拿在手中放入烤箱上层（视频约 460-545 s）、下层（545-637 s），后连托盘放入烤箱上层并关门加热约 750-963 s，外观无变化，未见熟的迹象。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：右夹爪夹住派送入烤箱腔体；后续夹托盘"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在烤箱前移动"
+     }
+    ]
+   },
+   "pie_on_tray": {
+    "method": "visual",
+    "intervals": [
+     [
+      687,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 687,
+      "log_step": 589,
+      "text_zh": "观察：视频 687-689 s 派放在台面深色托盘上，夹爪松开；agent 随后端起托盘（派仍在上）放入烤箱，视频 900 s、1093 s 派仍在托盘上。推断：q_score 0.5 与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：右夹爪夹派移到托盘上方放下，之后夹托盘边缘搬运"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驱动携带派至台面"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.5,
+   "unit": "ratio",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "pie_cooked",
+   "where": "操作中",
+   "t": 460,
+   "evidence_zh": "观察：284 s 掉落的派已在约 300-480 s 重新抓起，属已修复，不是断点。断点为约 460 s 首次送入烤箱之后：派在烤箱各腔体和连托盘的加热共约 600 s 仍未烹熟，视频末帧派在烤箱内托盘上，1081 s 仍在尝试取出托盘；未烹熟的原因无证据。",
+   "log_step": 460
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "未提供状态日志，无法核实派的温度与 Cooked 状态；派未熟的原因（烤箱类型是否能烤熟，参考运行用平顶烤炉）无证据，现象与根因均留空。",
+  "unmet": {
+   "pie_cooked": false,
+   "pie_on_tray": true
+  }
+ },
+ "summary_zh": "观察：打开冰箱（视频 44 s）、开启烤箱热源（258 s）、派放上托盘（687 s）达成，苹果派未烹熟，成功 0、q_score 0.5。断点在“苹果派烤熟”：派曾在 284 s 掉地，之后在烤箱各腔体和托盘上长时间加热仍无熟化迹象。缺口：无状态日志，派温度不可知，根因无证据。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点值（冰箱44 s、热源258 s、托盘687 s、烤熟未达成）与帧和q_score一致，未发现节点错误；断点证据把已修复的掉落当作断点锚点，gap_zh自相矛盾，需小改。",
+    "n_issues": 2,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点值（冰箱44 s、热源约258-270 s、托盘687 s、烤熟未达成）与帧及q_score=0.5一致；断点在pie_cooked、现象/根因留空并有gap_zh，证据不足以支持任何具体现象，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:05Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-cook-a-frozen-pie-i00-standard--ntsulhv/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-a-frozen-pie-i00-standard--ntsulhv/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-a-frozen-pie-i00-standard--ntsulhv/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-cook-a-frozen-pie-i00-standard--ntsulhv/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "1d4d1325152076ea3f5d4a0081c589b422f8c77232c2b8b405d7c73e0f1634c0",
+   "video_sha256": "8ecff6936c988eb5e3ab1b1ea1d1a3fde8967b175027f458a37133cf196ebdc4",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/cook_bacon/trajectories/behavior-cook-bacon-i00-standard--ewvjjkv.json b/phaseR/behavior-1k/cook_bacon/trajectories/behavior-cook-bacon-i00-standard--ewvjjkv.json
new file mode 100644
index 0000000000000000000000000000000000000000..345e171f7f84cc084b3bce7aeb92085aef85e9c7
--- /dev/null
+++ b/phaseR/behavior-1k/cook_bacon/trajectories/behavior-cook-bacon-i00-standard--ewvjjkv.json
@@ -0,0 +1,344 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-cook-bacon-i00-standard--ewvjjkv",
+ "taskKey": "behavior-1k/cook_bacon",
+ "dag_sha256": "436960e028055689855df9a0ab19653897ca4b01a7cdc6525e18a38048ea1e94",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 3759,
+   "sim_steps": 3909,
+   "physics_steps": 15636,
+   "steps": 621,
+   "elapsed_s": 6117.8,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 150,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.0,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.296,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.291,
+   "peak_cpus": 1.17,
+   "peak_pids": 151,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-cook-bacon-i00-standard--ewvjjkv.mp4",
+  "duration_s": 130.4,
+  "fps": 10.0,
+  "frames": 1304,
+  "sha256": "fe59b9bd9d22b2e5db2f0526307cd19b8c00f201a3150ff9d4654dae596deafa"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 9,
+  "base": 1,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      29.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 30.0,
+      "log_step": null,
+      "text_zh": "观察：视频 18-26 s 右侧视图中冰箱门仍贴合柜体，视频 30 s 门板已向外转开；之后至视频 130.4 s 均可见打开的冰箱和空架。推断：打开时刻约 28-30 s，不确定约 ±2 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "勾",
+      "note_zh": "观察：视频 30-36 s 机器人一只夹爪在门边缘/把手处拉门（视角较小，具体抓握方式未能确认）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驶到冰箱前"
+     }
+    ]
+   },
+   "heat_enabled": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程未见机器人触碰灶台按钮或旋钮；视频 30 s 灶台旋钮处无亮灯，q_score=0。"
+     }
+    ]
+   },
+   "bacon_cooked_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 130.0,
+      "log_step": null,
+      "text_zh": "观察：视频末尾培根未见到被放到热源上；评分器 q_score=0。推断：无培根达到熟状态。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频约 82-112 s 左夹爪夹托盘边缘取出（右臂顶住冰箱门）；约 115 s 后托盘落地，130 s 不再被夹持。"
+     }
+    ]
+   },
+   "bacon_cooked_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 130.0,
+      "log_step": null,
+      "text_zh": "观察：视频末尾培根未见到被放到热源上；评分器 q_score=0。推断：无培根达到熟状态。"
+     }
+    ]
+   },
+   "bacon_cooked_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 130.0,
+      "log_step": null,
+      "text_zh": "观察：视频末尾培根未见到被放到热源上；评分器 q_score=0。推断：无培根达到熟状态。"
+     }
+    ]
+   },
+   "bacon_cooked_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 130.0,
+      "log_step": null,
+      "text_zh": "观察：视频末尾培根未见到被放到热源上；评分器 q_score=0。推断：无培根达到熟状态。"
+     }
+    ]
+   },
+   "bacon_cooked_5": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 130.0,
+      "log_step": null,
+      "text_zh": "观察：视频末尾培根未见到被放到热源上；评分器 q_score=0。推断：无培根达到熟状态。"
+     }
+    ]
+   },
+   "bacon_cooked_6": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 130.0,
+      "log_step": null,
+      "text_zh": "观察：视频末尾培根未见到被放到热源上；评分器 q_score=0。推断：无培根达到熟状态。"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      29.0
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 125.0,
+      "log_step": null,
+      "text_zh": "观察：冰箱初始关闭（视频 0 起），约 29 s 被打开，视频 125-130.4 s 仍可见门敞开、架子外露，未再关闭。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "ratio",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bacon_cooked_1",
+   "t": 112.0,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 30 s 冰箱已打开；约 82-112 s 左夹爪夹托盘边缘取出，112 s 托盘被拉出冰箱，机器人始终持托盘/培根在冰箱旁操作，未把培根送到热源、未开热源；q_score=0。agent 在 log step 285-298 仍在整理托盘、拾取剩余培根。",
+   "log_step": 298
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 118.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 112 s 托盘被拉出冰箱后倾斜；118 s 中视图仍见托盘带培根，约 124-130 s 左右腕视图仅见地面上零散红色培根、冰箱架已空，托盘/培根脱离夹持落地（落地瞬间画面小、不清）。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "evidence_zh": "推断：左夹爪只夹托盘边缘单点取出，撤离时托盘倾斜后落地（视频 112-124 s）；agent 读数 log step 242/278 的 tcp 姿态变化仅作旁证。是否有可行的更稳抓法未验证。",
+    "source_zh": "视频"
+   }
+  ],
+  "gap_zh": "托盘落地的具体瞬间、落地后托盘与培根的细节在三视图小画面中不清；根因为推断，替代抓法未验证。agent 在 log step 68 遇到 Gemini 过滤器拦截但会话继续，未交付熟培根。",
+  "unmet": {
+   "bacon_cooked_1": false,
+   "bacon_cooked_2": false,
+   "bacon_cooked_3": false,
+   "bacon_cooked_4": false,
+   "bacon_cooked_5": false,
+   "bacon_cooked_6": false,
+   "fridge_closed": false
+  }
+ },
+ "summary_zh": "观察：录像（130.4 s）中机器人约 29 s 打开冰箱，约 82-112 s 单手夹托盘边缘取出培根，托盘倾斜后落地、培根散落地面（没保持住，操作中）；未开热源、无培根煮熟（q_score=0），冰箱未关。根因为推断（抓取和接触设计），画面小，置信低。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点值（冰箱约29 s打开、热源未开、无培根熟、q=0）与视频一致，断点ref取bacon_cooked_1可接受。但失败位置写成“转场”不对（机器人持托盘/培根在取出途中，属操作中）；且视频清楚显示托盘在约112 s被拉出冰箱后倾斜落到地面、培根散落地面，现象/根因被留空并以“视角受限”为由，证据其实支持“没保持住”。",
+    "n_issues": 4,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点值（冰箱约29 s打开、热源未开、无培根熟、q=0）、断点（bacon_cooked_1，操作中）、没保持住/抓取和接触设计（推断，已标低置信）与视频一致，第一轮意见已采纳，未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:43Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-cook-bacon-i00-standard--ewvjjkv/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-bacon-i00-standard--ewvjjkv/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-bacon-i00-standard--ewvjjkv/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-cook-bacon-i00-standard--ewvjjkv/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "d89e5fd960b403df72d10d2670a1754ecce213f0a47b46ed087bd58a5df9c327",
+   "video_sha256": "fe59b9bd9d22b2e5db2f0526307cd19b8c00f201a3150ff9d4654dae596deafa",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/cook_brussels_sprouts/trajectories/behavior-cook-brussels-sprouts-i00-standard--daxmjj7.json b/phaseR/behavior-1k/cook_brussels_sprouts/trajectories/behavior-cook-brussels-sprouts-i00-standard--daxmjj7.json
new file mode 100644
index 0000000000000000000000000000000000000000..dadc150ead29f119dfe1f298a5f6cc3bd7c32101
--- /dev/null
+++ b/phaseR/behavior-1k/cook_brussels_sprouts/trajectories/behavior-cook-brussels-sprouts-i00-standard--daxmjj7.json
@@ -0,0 +1,563 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-cook-brussels-sprouts-i00-standard--daxmjj7",
+ "taskKey": "behavior-1k/cook_brussels_sprouts",
+ "dag_sha256": "bf76cda0e3966e940a07729b29828c06815c78a9533d4989ef80a262c4504c19",
+ "outcome": {
+  "success": 0,
+  "reward": 0.07692307692307693,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 18549,
+   "sim_steps": 18699,
+   "physics_steps": 74796,
+   "steps": 17467,
+   "elapsed_s": 6864.4,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 150,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.07692307692307693,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "peak_cpus": 0.0,
+   "peak_pids": 0.0,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-cook-brussels-sprouts-i00-standard--daxmjj7.mp4",
+  "duration_s": 623.4,
+  "fps": 10.0,
+  "frames": 6234,
+  "sha256": "44407419a789ec8ebfe7d1ecc1875d0ca89c49a92cabce8ab5b6ac08c43d4e09"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 24,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      25.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 25.0,
+      "log_step": 102,
+      "text_zh": "观察：视频约 24-26 s 起头部相机可见冰箱门打开、露出内部搁架与保鲜盒，30 s 门大开（agent 自述 log step 102 门已开）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：夹爪抓住左冰箱门外缘小凸起并拉开（log step 92-102）；底盘移动至门前。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驶至冰箱前"
+     }
+    ]
+   },
+   "oven_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程未见机器人接近或打开烤箱的画面；agent 自述（log step 305）仅称“之后再处理烤箱”，从未做。"
+     }
+    ]
+   },
+   "sprouts_halved_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 263,
+      "text_zh": "观察：约 330-400 s 前后画面可见台面上一颗绿色小块，agent 自述切了一刀得到一个半颗（log step 263，自述）；评分器观察：outcome.raw.q_score=1/13，已满足的 1 项对应哪个字面量未知，不据此断言半颗数量。本节点需要至少2个半颗同时存在，画面与日志均不能证明，故未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "压/擦",
+      "note_zh": "观察：agent 自述夹爪持切肉刀在台面上对一颗甘蓝切了一刀（log step 250）；画面未能逐帧确认。"
+     }
+    ]
+   },
+   "sprouts_halved_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "sprouts_halved_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "sprouts_halved_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_in_oven_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_in_oven_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_in_oven_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_in_oven_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_in_oven_5": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_in_oven_6": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_in_oven_7": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_in_oven_8": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_cooked_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_cooked_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_cooked_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_cooked_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_cooked_5": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_cooked_6": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_cooked_7": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "halves_cooked_8": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 318,
+      "text_zh": "观察：冰箱开局关闭，约 25 s 被打开；agent 自述 log step 318“门似乎已关上”，但视频末段画面无法确认门是否关闭（门缝/开度不可见），故未知。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "agent 自述推门（log step 313-318），视频无法确认结果"
+     }
+    ]
+   },
+   "oven_closed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：全程视频（头部相机，视角频繁晃动）未见该步骤成立的画面；仅据画面判断。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.07692307692307693,
+   "unit": "1",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "sprouts_halved_1",
+   "t": 260.0,
+   "where": "操作中",
+   "evidence_zh": "观察：agent 取出保鲜盒后甘蓝在搬运/放置时洒落（log step 161），回收花了很长时间（log step 161-235）；之后切第一颗，agent 自述持刀抓取失败（log step 275-293，画面未能确认），其余两颗始终未切。第1颗节点（至少2个半颗）未能确认达成，故断点定在首个未达成的 sprouts_halved_1（取料后的切割与持刀操作）。",
+   "log_step": 250
+  },
+  "phenomena": [],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "evidence_zh": "推断：保鲜盒用边缘抓握搬运时甘蓝洒落（log step 161），刀的抓握也未保持（log step 275），均是抓取方式不稳；无其余证据区分更细的子类。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "无状态日志；头部相机视角频繁晃动，切割、半颗数量、冰箱最终开闭状态均无法在画面确认；现象/根因主要依据 agent 日志自述，故较弱。 “刀脱手/落地”仅有 agent 自述（log step 275-293），280-300 s 画面（晃动）未见刀落地，故不列接触现象。",
+  "unmet": {
+   "sprouts_halved_1": false,
+   "sprouts_halved_2": false,
+   "sprouts_halved_3": false,
+   "sprouts_halved_4": false,
+   "halves_in_oven_1": false,
+   "halves_in_oven_2": false,
+   "halves_in_oven_3": false,
+   "halves_in_oven_4": false,
+   "halves_in_oven_5": false,
+   "halves_in_oven_6": false,
+   "halves_in_oven_7": false,
+   "halves_in_oven_8": false,
+   "halves_cooked_1": false,
+   "halves_cooked_2": false,
+   "halves_cooked_3": false,
+   "halves_cooked_4": false,
+   "halves_cooked_5": false,
+   "halves_cooked_6": false,
+   "halves_cooked_7": false,
+   "halves_cooked_8": false,
+   "fridge_closed": null,
+   "oven_closed": false
+  }
+ },
+ "summary_zh": "观察：机器人在视频约 25 s 打开冰箱，取出保鲜盒后甘蓝洒落，花大量时间回收，只对一颗甘蓝切了一刀（agent 自述；评分器 q_score=1/13，对应字面量未知），随后据自述持刀失败，烤箱从未打开，八个半颗均未入炉或烤熟。断点在第1个切半节点（持刀切割操作）。证据缺口：无状态日志，画面难以确认半颗数量与冰箱最终是否关闭。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 3,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 5,
+    "summary_zh": "节点主体（冰箱开、其余未达成）大体可接受，但断点与节点不自洽，且把评分器读数误称为 agent 自述。头部相机画面剧烈晃动，切割与半颗数量无法由画面判定。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "冰箱开启时间标错：视频约 24-26 s 起头部相机已可见冰箱门打开并露出保鲜盒，而标注为 45 s（晚约 20 s）。另外接触类现象仅依据 agent 自述。其余（烤箱及后续节点未达成、断点定在 sprouts_halved_1）与评分器 q=1/13 及画面相容。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 3,
+    "kind": "confirm",
+    "verdict": "agree",
+    "summary_zh": "I1、I2 两项接受的修改均已正确应用，与画面相符。",
+    "n_issues": 0,
+    "n_checks": 2
+   }
+  ],
+  "open_zh": [],
+  "confirm": {
+   "verdict": "confirmed",
+   "summary_zh": "I1、I2 两项接受的修改均已正确应用，与画面相符。",
+   "checks_ok": "2/2",
+   "n_factual_errors": 0
+  }
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:53Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--daxmjj7/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--daxmjj7/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--daxmjj7/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--daxmjj7/04_review_r2"
+   },
+   {
+    "stage": "reconcile",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--daxmjj7/05_reconcile_r2"
+   },
+   {
+    "stage": "confirm",
+    "round": 3,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--daxmjj7/06_confirm_r3"
+   }
+  ],
+  "confirm_prompt": "prompts/06_confirm.md",
+  "confirm_prompt_sha256": "b9f56c095f7ea5e609e58bb21e076f565766c370333d71c7a3512cffb4d78267",
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "7533f600759b1b7a59403b96602ebd6c5e6c0fad183d4a7e9834f08579dc6732",
+   "video_sha256": "44407419a789ec8ebfe7d1ecc1875d0ca89c49a92cabce8ab5b6ac08c43d4e09",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/cook_brussels_sprouts/trajectories/behavior-cook-brussels-sprouts-i00-standard--kvrigdf.json b/phaseR/behavior-1k/cook_brussels_sprouts/trajectories/behavior-cook-brussels-sprouts-i00-standard--kvrigdf.json
new file mode 100644
index 0000000000000000000000000000000000000000..729fda8c7637b7c9f051f4cbff990699f84cd5e6
--- /dev/null
+++ b/phaseR/behavior-1k/cook_brussels_sprouts/trajectories/behavior-cook-brussels-sprouts-i00-standard--kvrigdf.json
@@ -0,0 +1,604 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-cook-brussels-sprouts-i00-standard--kvrigdf",
+ "taskKey": "behavior-1k/cook_brussels_sprouts",
+ "dag_sha256": "bf76cda0e3966e940a07729b29828c06815c78a9533d4989ef80a262c4504c19",
+ "outcome": {
+  "success": 0,
+  "reward": 0.2692307692307692,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 29527,
+   "sim_steps": 29527,
+   "physics_steps": 118108,
+   "steps": 19049,
+   "elapsed_s": 11085.5,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "replay_error": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "q_score": 0.2692307692307692,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.857,
+   "peak_processes": 55,
+   "peak_disk_gb": 0.672,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-cook-brussels-sprouts-i00-standard--kvrigdf/replay.mp4",
+  "duration_s": 984.234,
+  "fps": 30.0,
+  "frames": 29527,
+  "sha256": "d91e4d2add627b90b079d19cec737f85df5946dc9ff8319ed146a1521c321ffa"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 24,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      62.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 62.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 60-130 s 头部相机可见冰箱双门敞开、露出内部层架；之后机器人从冰箱取出保鲜盒（约 150 s 起盒内可见甘蓝）。开门时刻约 62 s（不确定 ±5 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "机器人夹爪握住冰箱门把手/边缘拉开；底盘移动到门前"
+     }
+    ]
+   },
+   "oven_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      405.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 405.0,
+      "log_step": 197,
+      "text_zh": "观察：视频 350-400 s 机器人多次拉/推烤箱门，门反复被推回；约 405-415 s 起门板大幅张开露出腔体，终盘（约 900-960 s）烤箱门仍张开。时刻约 405 s（350 s 处已略开又被推回，不确定）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "视频 350-415 s 机器人手臂拉或推烤箱门（含多次被推回）；底盘移动调整位置"
+     }
+    ]
+   },
+   "sprouts_halved_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 250.0,
+      "log_step": null,
+      "text_zh": "观察：视频 190-300 s 机器人持刀在保鲜盒/砧板处切甘蓝，但头部及腕部相机分辨率不足，无法数出整颗被两个半颗替代。推断：评分总 q=7/26 允许至多 7 个半颗 real；agent 自述（log step 166）已切完三颗，仅作会话证据，不能当节点值。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "压/擦",
+      "note_zh": "机器人夹爪持刀在保鲜盒上方切甘蓝"
+     }
+    ]
+   },
+   "sprouts_halved_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 250.0,
+      "log_step": null,
+      "text_zh": "观察：视频 190-300 s 机器人持刀在保鲜盒/砧板处切甘蓝，但头部及腕部相机分辨率不足，无法数出整颗被两个半颗替代。推断：评分总 q=7/26 允许至多 7 个半颗 real；agent 自述（log step 166）已切完三颗，仅作会话证据，不能当节点值。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "压/擦",
+      "note_zh": "机器人夹爪持刀在保鲜盒上方切甘蓝"
+     }
+    ]
+   },
+   "sprouts_halved_3": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 250.0,
+      "log_step": null,
+      "text_zh": "观察：视频 190-300 s 机器人持刀在保鲜盒/砧板处切甘蓝，但头部及腕部相机分辨率不足，无法数出整颗被两个半颗替代。推断：评分总 q=7/26 允许至多 7 个半颗 real；agent 自述（log step 166）已切完三颗，仅作会话证据，不能当节点值。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "压/擦",
+      "note_zh": "机器人夹爪持刀在保鲜盒上方切甘蓝"
+     }
+    ]
+   },
+   "sprouts_halved_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 300.0,
+      "log_step": null,
+      "text_zh": "推断：评分器 q_score=7/26（0.269），若第4颗切半成立则 8 个半颗 real 已占 8 条文字，加上其余目标不可能只得 7，故 final 为 false（逻辑推断自总分，CASES PROG-10）。agent 自述（log step 174）称第四颗已分开，与总分不符。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "压/擦",
+      "note_zh": "机器人夹爪持刀在保鲜盒上方切甘蓝"
+     }
+    ]
+   },
+   "halves_in_oven_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 900.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 880-940 s 烤箱门张开，机器人双手携半颗接近；腕部/头部视角看不清炉腔内的半颗数量。推断（逻辑，推断自总分，CASES PROG-10）：q_score=7/26，每个真实半颗、每个在炉内、每个熟的半颗各占一条文字；入炉k个至少需k个真实半颗，4+4=8>7，故入炉数≤3。入炉数在0-3之间，具体几个无法确定，故 unknown；agent 自述（log step 345）仅作会话叙述，不作证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "双夹爪各抓一个半颗，底盘驱动携带到烤箱放入腔底"
+     }
+    ]
+   },
+   "halves_in_oven_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 900.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 880-940 s 烤箱门张开，机器人双手携半颗接近；腕部/头部视角看不清炉腔内的半颗数量。推断（逻辑，推断自总分，CASES PROG-10）：q_score=7/26，每个真实半颗、每个在炉内、每个熟的半颗各占一条文字；入炉k个至少需k个真实半颗，4+4=8>7，故入炉数≤3。入炉数在0-3之间，具体几个无法确定，故 unknown；agent 自述（log step 345）仅作会话叙述，不作证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "双夹爪各抓一个半颗，底盘驱动携带到烤箱放入腔底"
+     }
+    ]
+   },
+   "halves_in_oven_3": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 900.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 880-940 s 烤箱门张开，机器人双手携半颗接近；腕部/头部视角看不清炉腔内的半颗数量。推断（逻辑，推断自总分，CASES PROG-10）：q_score=7/26，每个真实半颗、每个在炉内、每个熟的半颗各占一条文字；入炉k个至少需k个真实半颗，4+4=8>7，故入炉数≤3。入炉数在0-3之间，具体几个无法确定，故 unknown；agent 自述（log step 345）仅作会话叙述，不作证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "双夹爪各抓一个半颗，底盘驱动携带到烤箱放入腔底"
+     }
+    ]
+   },
+   "halves_in_oven_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 900.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 880-940 s 烤箱门张开，机器人携半颗往返烤箱，看不清炉内。推断（逻辑，推断自总分，CASES PROG-10）：q_score=7/26，每个真实半颗、每个在炉内、每个熟的半颗各占一条文字；入炉k个至少需k个真实半颗，4+4=8>7，故入炉数≤3。故第4个半颗入炉 final false。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "双夹爪各抓一个半颗，底盘驱动携带到烤箱放入腔底"
+     }
+    ]
+   },
+   "halves_in_oven_5": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频末段 984 s 烤箱门仍张开。推断（逻辑，推断自总分，CASES PROG-10）：q_score=7/26，每个真实半颗、每个在炉内、每个熟的半颗各占一条文字；入炉k个至少需k个真实半颗，4+4=8>7，故入炉数≤3。故第5个半颗入炉 final false；不依赖 agent 自述。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪尝试抓取地面/保鲜盒中的半颗（含两次抓空），未送入炉内"
+     }
+    ]
+   },
+   "halves_in_oven_6": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频末段烤箱门仍张开，机器人未再往炉内放半颗；agent 自述（log step 345）称仅四个半颗入炉，第5-8个未入炉。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪尝试抓取地面/保鲜盒中的半颗（含两次抓空），未送入炉内"
+     }
+    ]
+   },
+   "halves_in_oven_7": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频末段烤箱门仍张开，机器人未再往炉内放半颗；agent 自述（log step 345）称仅四个半颗入炉，第5-8个未入炉。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪尝试抓取地面/保鲜盒中的半颗（含两次抓空），未送入炉内"
+     }
+    ]
+   },
+   "halves_in_oven_8": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频末段烤箱门仍张开，机器人未再往炉内放半颗；agent 自述（log step 345）称仅四个半颗入炉，第5-8个未入炉。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪尝试抓取地面/保鲜盒中的半颗（含两次抓空），未送入炉内"
+     }
+    ]
+   },
+   "halves_cooked_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频无法显示熟态变化（无参考视频、相机分辨率不足）；q 总分不能单独分配各条熟态。"
+     }
+    ]
+   },
+   "halves_cooked_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频无法显示熟态变化（无参考视频、相机分辨率不足）；q 总分不能单独分配各条熟态。"
+     }
+    ]
+   },
+   "halves_cooked_3": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频无法显示熟态变化（无参考视频、相机分辨率不足）；q 总分不能单独分配各条熟态。"
+     }
+    ]
+   },
+   "halves_cooked_4": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频无法显示熟态变化（无参考视频、相机分辨率不足）；q 总分不能单独分配各条熟态。"
+     }
+    ]
+   },
+   "halves_cooked_5": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频无法显示熟态变化（无参考视频、相机分辨率不足）；q 总分不能单独分配各条熟态。"
+     }
+    ]
+   },
+   "halves_cooked_6": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频无法显示熟态变化（无参考视频、相机分辨率不足）；q 总分不能单独分配各条熟态。"
+     }
+    ]
+   },
+   "halves_cooked_7": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频无法显示熟态变化（无参考视频、相机分辨率不足）；q 总分不能单独分配各条熟态。"
+     }
+    ]
+   },
+   "halves_cooked_8": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：视频无法显示熟态变化（无参考视频、相机分辨率不足）；q 总分不能单独分配各条熟态。"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0.0,
+      62.0
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：冰箱开局关闭，视频约 62 s 被打开；末段（约 970 s）头部相机仍见冰箱双门敞开，未再关上。"
+     }
+    ]
+   },
+   "oven_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0.0,
+      350.0
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 984.0,
+      "log_step": null,
+      "text_zh": "观察：烤箱开局关闭；视频 350 s 起被拉开；末段 900-984 s 烤箱门仍张开。agent 自述 log step 367 称门约 3-4° 在推关，但终盘画面显示门板张开。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "约 930-984 s 机器人手臂推烤箱门，终盘门仍张开"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.2692307692307692,
+   "unit": "1",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "sprouts_halved_4",
+   "t": 290.0,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "推断（逻辑，推断自总分，CASES PROG-10）：q_score=7/26，每个真实半颗占一条文字，至多7个半颗存在，第4颗切半（8个半颗）final false；炉内无法确认有半颗，故第一个未补救的未完成步是切割，对应视频约 282-292 s 持刀切甘蓝段。观察：该段机器人持刀在保鲜盒处切甘蓝，分辨率不足以区分切过头与未切开，故不写现象（见 gap_zh）。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "头部与腕部相机分辨率低，看不出半颗位置/掉落细节；无状态日志，故无法确定具体接触失败现象与根因，agent 自述（log step 301 保鲜盒翻倒、343 抓空两次）仅为会话叙述，不作现象。token 预算耗尽（log step 373）本身不是现象。视频为 984 s，与 agent 11000 s 会话时钟无对应，未换算。第1-3颗切半与入炉1-3因无法目视计数且总分只给上限，标 unknown；第4颗切半及入炉4-8 final false 推断自总分q=7/26；切割段无法区分切过头与未切开。",
+  "unmet": {
+   "sprouts_halved_1": null,
+   "sprouts_halved_2": null,
+   "sprouts_halved_3": null,
+   "sprouts_halved_4": false,
+   "halves_in_oven_1": null,
+   "halves_in_oven_2": null,
+   "halves_in_oven_3": null,
+   "halves_in_oven_4": false,
+   "halves_in_oven_5": false,
+   "halves_in_oven_6": false,
+   "halves_in_oven_7": false,
+   "halves_in_oven_8": false,
+   "halves_cooked_1": null,
+   "halves_cooked_2": null,
+   "halves_cooked_3": null,
+   "halves_cooked_4": null,
+   "halves_cooked_5": null,
+   "halves_cooked_6": null,
+   "halves_cooked_7": null,
+   "halves_cooked_8": null,
+   "fridge_closed": false,
+   "oven_closed": false
+  }
+ },
+ "summary_zh": "观察：视频 984 s 内机器人打开冰箱（约 62 s）、取出保鲜盒、持刀切甘蓝、反复拉推烤箱门至张开（约 405 s），并多次携半颗往返烤箱；q_score=7/26，终盘烤箱和冰箱门均张开。推断：第4颗切半 final false（总分上限），第4-8个半颗入炉及两扇关门未达成（入炉数≤3，推断自总分），断点在第4颗切半（约282-292 s）。第1-3颗切半、入炉1-3和烤熟无法目视计数，标 unknown，现象与根因无证据留空。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "断点选错：q=7/26（每个真实半颗一条文字）已使第4颗切半 final=false，且「入炉4个半颗」与总分矛盾（4个入炉需≥4个real，4+4=8>7），故 halves_in_oven_4 也应为 false；按 CASES PROG-10，第一个未补救步骤是第4颗切半，而非第5个半颗入炉。",
+    "n_issues": 3,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "第1轮问题已修正：断点为 sprouts_halved_4（推断自 q=7/26，PROG-10），入炉4-8为 false，其余无法判定者 unknown，均有总分依据；现象/根因为空且有 gap_zh，因分辨率不足以区分失败方式，可接受。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:09Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--kvrigdf/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--kvrigdf/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--kvrigdf/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-cook-brussels-sprouts-i00-standard--kvrigdf/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "94f0ea58f14be354e95286fb35211d9d05eb479e233da773fc5abd8857c3022e",
+   "video_sha256": "d91e4d2add627b90b079d19cec737f85df5946dc9ff8319ed146a1521c321ffa",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/cook_cabbage/trajectories/behavior-cook-cabbage-i00-standard--wwazavk.json b/phaseR/behavior-1k/cook_cabbage/trajectories/behavior-cook-cabbage-i00-standard--wwazavk.json
new file mode 100644
index 0000000000000000000000000000000000000000..969d3cadde7b117edd9155aa85d36fafb3e8894f
--- /dev/null
+++ b/phaseR/behavior-1k/cook_cabbage/trajectories/behavior-cook-cabbage-i00-standard--wwazavk.json
@@ -0,0 +1,316 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-cook-cabbage-i00-standard--wwazavk",
+ "taskKey": "behavior-1k/cook_cabbage",
+ "dag_sha256": "6c041940456d17727dc9db527f9c928aaf9fc5d2815fc6bdacfa60aff9b851bd",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 4647,
+   "sim_steps": 4797,
+   "physics_steps": 19188,
+   "steps": 761,
+   "elapsed_s": 6195.2,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 150,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.0,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.275,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.282,
+   "peak_cpus": 1.09,
+   "peak_pids": 150,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-cook-cabbage-i00-standard--wwazavk.mp4",
+  "duration_s": 160.0,
+  "fps": 10.0,
+  "frames": 1600,
+  "sha256": "1444b19dd4a7207bb2a5ad2a397df661917547189bc9d64413e420d66beb9a83"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 7,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      40.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 40.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 40-44 s 起，头/腕部相机可见冰箱门已离开箱体，隔层内可见卷心菜和辣椒；之后 48-76 s 机器人从冰箱取出辣椒和卷心菜。时间约±3 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "agent 多次尝试用夹爪抓/勾/推把手开门（log step 57-88 的脚本叙述），视频只能确认门最终打开，具体接触方式未逐帧核实"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动到冰箱前"
+     }
+    ]
+   },
+   "heat_source_on": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 159.5,
+      "log_step": null,
+      "text_zh": "观察：视频 96-160 s 看到灶具，灶眼无火焰、无绿色指示；机器人从未触碰灶具旋钮。"
+     }
+    ]
+   },
+   "pan_at_heat_source": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 159.5,
+      "log_step": null,
+      "text_zh": "观察：视频 108-160 s 煎锅一直在灶具左侧的台面上，锅底不在灶眼上；机器人只在锅内放辣椒、切菜，没有搬锅。"
+     }
+    ]
+   },
+   "cabbage_diced": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 100.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 100 s 起地面可见多颗绿色圆球（卷心菜）掉在地上，没有被切成丁；之后右手拿刀切的是锅里的辣椒。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "一侧腕部夹爪（左右未核实，疑为左）曾抓着卷心菜（视频 56-100 s 腕部相机），随后卷心菜落到地上"
+     }
+    ]
+   },
+   "chili_diced": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 150.0,
+      "log_step": null,
+      "text_zh": "观察：视频 125-160 s 右夹爪握刀在煎锅上方动作，锅内可见红色小块，但分辨率不足以判断是否已切成丁；agent 最后一步仍在启动切辣椒脚本（log step 223）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右夹爪抓住菜刀在锅内辣椒上切（视频约 125-160 s）"
+     }
+    ]
+   },
+   "cabbage_cooked_in_pan": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 159.5,
+      "log_step": null,
+      "text_zh": "观察：视频结束时锅内无卷心菜，卷心菜在地上；热源未开。评分 q_score=0。"
+     }
+    ]
+   },
+   "chili_cooked_in_pan": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 159.5,
+      "log_step": null,
+      "text_zh": "观察：辣椒放进了锅（约 100 s），但热源未开，没有烹熟；q_score=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "一侧腕部夹爪（与抓卷心菜的手相对的另一只，具体左右未核实）抓着辣椒搬到锅里放下"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "比例",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cabbage_diced",
+   "t": 100.0,
+   "log_step": null,
+   "where": "转场",
+   "evidence_zh": "观察：视频约 56-100 s 机器人一手抓卷心菜、另一手抓辣椒，搬运途中卷心菜落到地面（约 100 s 起地上可见绿球），之后只处理辣椒，卷心菜从未被捡回、切丁。推断：卷心菜切丁一步在搬运中失败且未修复。"
+  },
+  "unmet": {
+   "cabbage_cooked_in_pan": false,
+   "chili_cooked_in_pan": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 100.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 56-96 s 腕部相机显示卷心菜被夹爪抓住，约 100 s 后地面出现卷心菜，说明搬运中丢失。具体掉落瞬间未逐帧核实。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "根因不明：视频未清晰显示掉落原因（抓取过紧/过松、碰撞），无状态日志，agent 日志无对应叙述，故根因留空。另：agent 最后阶段仍在写切菜脚本，热源未开、锅未上灶，后续步骤同样未完成，但按未修复的最早步骤取卷心菜切丁。冰箱开启时间 ±3 s，辣椒切丁是否完成无法判断。"
+ },
+ "summary_zh": "观察：视频 160 s 内机器人开了冰箱（约 40 s），取出辣椒和卷心菜，辣椒放进锅内，右手持刀在锅上切（约 125-160 s）；卷心菜约 100 s 起落在地面，灶具未开、锅未放上灶眼，q_score=0。推断：断点在卷心菜切丁（搬运中丢失），目标节点均未达成；缺口：无状态日志，掉落原因与辣椒是否切丁无法判定。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "主要标签（冰箱约 40-46 s 开启、卷心菜约 98-102 s 搬运中掉落、断点 cabbage_diced/转场、没保持住、根因留空并有 gap）与视频一致；仅 chili_cooked_in_pan 的 contact_actual 把辣椒也写成左腕夹爪，与同文件 cabbage_diced 的左腕抓卷心菜及断点文字『另一手抓辣椒』自相矛盾。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "冰箱约40 s开启、卷心菜约100 s落地、辣椒入锅、热源未开等标签与抽查帧一致；断点cabbage_diced/转场、没保持住、根因留空并带gap_zh均有据，未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:31Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-cook-cabbage-i00-standard--wwazavk/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-cabbage-i00-standard--wwazavk/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-cabbage-i00-standard--wwazavk/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-cook-cabbage-i00-standard--wwazavk/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "18585635fcbc78517727915513494226f795f1a61df5d98570bae398c2eeac26",
+   "video_sha256": "1444b19dd4a7207bb2a5ad2a397df661917547189bc9d64413e420d66beb9a83",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/cook_hot_dogs/trajectories/behavior-cook-hot-dogs-i00-standard--zqp8nx5.json b/phaseR/behavior-1k/cook_hot_dogs/trajectories/behavior-cook-hot-dogs-i00-standard--zqp8nx5.json
new file mode 100644
index 0000000000000000000000000000000000000000..84d07935f00bc97a246f2b607afff0e457235c52
--- /dev/null
+++ b/phaseR/behavior-1k/cook_hot_dogs/trajectories/behavior-cook-hot-dogs-i00-standard--zqp8nx5.json
@@ -0,0 +1,206 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-cook-hot-dogs-i00-standard--zqp8nx5",
+ "taskKey": "behavior-1k/cook_hot_dogs",
+ "dag_sha256": "e4e56cb15fd41bb6774ed8f95eaf0a9a5618338c0ee4fa13e4c93482b67b350d",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 35251,
+   "sim_steps": 35251,
+   "physics_steps": 141004,
+   "steps": 18547,
+   "elapsed_s": 13464.8,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "replay_error": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "q_score": 0.0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.316,
+   "peak_processes": 12,
+   "peak_disk_gb": 0.349,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-cook-hot-dogs-i00-standard--zqp8nx5/replay.mp4",
+  "duration_s": 1175.034,
+  "fps": 30.0,
+  "frames": 35251,
+  "sha256": "145bd4563c00f3b731be3ea0b46fd9e3f56d91096085d33d98c6fe7be353f951"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hotdogs_cooked_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1174,
+      "log_step": null,
+      "text_zh": "观察：视频 1174 s 末帧机器人仍在冰箱门旁，炉灶旁未见熟食；评分器 q_score=0（推断：无熟食文字）。视频约 750-775 s 左夹爪持一根热狗离开冰箱，约 790-805 s 放在灶台，约 815-840 s 按开关；约 900-1174 s 机器人回冰箱处理第二根。观察：开关标记约 830 s 一度变绿、约 840 s 回红；推断：炉灶曾短暂开启后又关闭或未持续加热；两根热狗均无熟度证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：左夹爪从冰箱玻璃架取出一根热狗并搬到灶台（视频约 750-860 s：取出约 750-775 s，放到灶台约 790-805 s；agent 自述 log step 302、314），随后用指尖按炉灶开关标记（约 815-840 s），开关标记约 830 s 变绿、840 s 回红。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驱动往返冰箱与灶台。"
+     }
+    ]
+   },
+   "hotdogs_cooked_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1174,
+      "log_step": null,
+      "text_zh": "观察：视频 1174 s 末帧机器人仍在冰箱门旁，炉灶旁未见熟食；评分器 q_score=0（推断：无熟食文字）。视频约 750-775 s 左夹爪持一根热狗离开冰箱，约 790-805 s 放在灶台，约 815-840 s 按开关；约 900-1174 s 机器人回冰箱处理第二根。观察：开关标记约 830 s 一度变绿、约 840 s 回红；推断：炉灶曾短暂开启后又关闭或未持续加热；两根热狗均无熟度证据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频末段左手推扫冰箱门内侧以取第二根热狗（agent 自述 log step 346-350），未取出。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "hotdogs_cooked_1",
+   "t": null,
+   "log_step": 314,
+   "where": "操作中",
+   "evidence_zh": "观察：热狗放到灶台后，机器人按炉灶开关（agent 自述 log step 314）；视频末帧无熟食迹象，q_score=0，仿真未判定任何热狗煮熟。观察：视频约 830 s 开关标记一度变绿、约 840 s 回红，热狗留在灶台。推断：炉灶曾短暂开启后又被关闭或未持续加热，熟度无证据。"
+  },
+  "unmet": {
+   "hotdogs_cooked_1": false,
+   "hotdogs_cooked_2": false
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "开关标记曾变绿后回红，但无状态日志，无法判定是被再次关闭还是加热不足。agent 因 token 预算耗尽（log step 357 API Error）而结束，不作为现象或根因。第二根热狗（hot dog A）一直未取出。"
+ },
+ "summary_zh": "观察：视频 1175 s 内机器人从冰箱取出一根热狗放到灶台并尝试按开关，随后回冰箱门处取第二根，但无热狗煮熟（q_score 0）。断点在第1根热狗煮熟（点炉灶失败）；现象与根因证据不足留空。主要缺口：无状态日志，炉灶状态不明。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "断点节点与 final 基本合理，但证据里的视频时间错了：取热狗、搬到灶台、按开关发生在视频约 750-860 s，而非 960-1000 s（960 s 后机器人已回到冰箱门处）；且炉灶开关标记在视频 ~830 s 变绿、~840 s 又变红，文件称“未见点火成功/无法确认”缺少这条可见线索，应如实写出。",
+    "n_issues": 2,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "第2轮修正后的标签与视频一致：热狗取出约750-775 s、放灶台、开关标记约830 s变绿840 s回红；无状态日志，q=0，现象与根因留空并有gap_zh，合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:38Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-cook-hot-dogs-i00-standard--zqp8nx5/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-hot-dogs-i00-standard--zqp8nx5/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-cook-hot-dogs-i00-standard--zqp8nx5/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-cook-hot-dogs-i00-standard--zqp8nx5/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "2d6d8acfa1b267b9288e52fc4f5a9f2334202cc19427aa3cde6a7c665b2a4d85",
+   "video_sha256": "145bd4563c00f3b731be3ea0b46fd9e3f56d91096085d33d98c6fe7be353f951",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/freeze_fruit/trajectories/behavior-freeze-fruit-i00-privileged--cxuoqcb.json b/phaseR/behavior-1k/freeze_fruit/trajectories/behavior-freeze-fruit-i00-privileged--cxuoqcb.json
new file mode 100644
index 0000000000000000000000000000000000000000..7aea7d1ca84ed76900c849a82242edb0edf8de54
--- /dev/null
+++ b/phaseR/behavior-1k/freeze_fruit/trajectories/behavior-freeze-fruit-i00-privileged--cxuoqcb.json
@@ -0,0 +1,302 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-freeze-fruit-i00-privileged--cxuoqcb",
+ "taskKey": "behavior-1k/freeze_fruit",
+ "dag_sha256": "e86f72e029e34ab3019ced83d0f54559848fd3255afc274eb31a7668669172af",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "deterministic": 0,
+   "missing_trajectory": 1,
+   "grader_error": 0,
+   "infra_error": 0,
+   "agent_sessions": 5,
+   "agent_resets": 1,
+   "agent_sim_steps": 30,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "none",
+  "url": null,
+  "duration_s": null,
+  "fps": null,
+  "frames": null,
+  "sha256": null
+ },
+ "state": null,
+ "progress": {
+  "method": "none",
+  "denominator": 9,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "cabinet_opened": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 502,
+      "text_zh": "观察：本次运行没有视频、没有状态日志，也没有交付轨迹（reward.missing_trajectory=1，agent_records=0）；log 最后一步（step 502）仍在写提取保鲜盒的测试脚本。该节点无法判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，无被评分的接触。"
+     }
+    ]
+   },
+   "fridge_opened": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 502,
+      "text_zh": "观察：本次运行没有视频、没有状态日志，也没有交付轨迹（reward.missing_trajectory=1，agent_records=0）；log 最后一步（step 502）仍在写提取保鲜盒的测试脚本。该节点无法判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，无被评分的接触。"
+     }
+    ]
+   },
+   "apples_in_boxes_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 502,
+      "text_zh": "观察：本次运行没有视频、没有状态日志，也没有交付轨迹（reward.missing_trajectory=1，agent_records=0）；log 最后一步（step 502）仍在写提取保鲜盒的测试脚本。该节点无法判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，无被评分的接触。"
+     }
+    ]
+   },
+   "apples_in_boxes_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 502,
+      "text_zh": "观察：本次运行没有视频、没有状态日志，也没有交付轨迹（reward.missing_trajectory=1，agent_records=0）；log 最后一步（step 502）仍在写提取保鲜盒的测试脚本。该节点无法判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，无被评分的接触。"
+     }
+    ]
+   },
+   "strawberries_in_boxes_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 502,
+      "text_zh": "观察：本次运行没有视频、没有状态日志，也没有交付轨迹（reward.missing_trajectory=1，agent_records=0）；log 最后一步（step 502）仍在写提取保鲜盒的测试脚本。该节点无法判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，无被评分的接触。"
+     }
+    ]
+   },
+   "strawberries_in_boxes_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 502,
+      "text_zh": "观察：本次运行没有视频、没有状态日志，也没有交付轨迹（reward.missing_trajectory=1，agent_records=0）；log 最后一步（step 502）仍在写提取保鲜盒的测试脚本。该节点无法判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，无被评分的接触。"
+     }
+    ]
+   },
+   "boxes_in_fridge_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 502,
+      "text_zh": "观察：本次运行没有视频、没有状态日志，也没有交付轨迹（reward.missing_trajectory=1，agent_records=0）；log 最后一步（step 502）仍在写提取保鲜盒的测试脚本。该节点无法判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，无被评分的接触。"
+     }
+    ]
+   },
+   "boxes_in_fridge_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 502,
+      "text_zh": "观察：本次运行没有视频、没有状态日志，也没有交付轨迹（reward.missing_trajectory=1，agent_records=0）；log 最后一步（step 502）仍在写提取保鲜盒的测试脚本。该节点无法判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，无被评分的接触。"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 502,
+      "text_zh": "观察：本次运行没有视频、没有状态日志，也没有交付轨迹（reward.missing_trajectory=1，agent_records=0）；log 最后一步（step 502）仍在写提取保鲜盒的测试脚本。该节点无法判断。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，无被评分的接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "cabinet_opened",
+   "t": null,
+   "log_step": 502,
+   "where": "首步前",
+   "evidence_zh": "观察：reward 中 missing_trajectory=1、agent_records=0、agent_replays=0，没有交付任何轨迹，graded 回放不存在；log 到 step 502（agent 工作约 6392 s）仍在调试打开橱柜门（门角度 q≈0.7）和提取保鲜盒的会话内测试。推断：没有被评分的操作，最先需要的是打开橱柜（取保鲜盒），故标在第一步前。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无视频、无状态日志，且未交付轨迹，没有可评分的操作过程；会话内测试仅是会话证据，不能作为失败现象或根因（agent 在预算内未交付属于 agent 侧原因，不贴“超时”标签）。所有节点 unknown。",
+  "unmet": {
+   "apples_in_boxes_1": null,
+   "apples_in_boxes_2": null,
+   "strawberries_in_boxes_1": null,
+   "strawberries_in_boxes_2": null,
+   "boxes_in_fridge_1": null,
+   "boxes_in_fridge_2": null,
+   "fridge_closed": null
+  }
+ },
+ "summary_zh": "观察：agent 在约 6400 s 的会话里一直在探索和调试橱柜开门与取盒的脚本（log step 3-502），没有保存任何轨迹（missing_trajectory=1）。推断：被评分的尝试为空，所有节点无法判断，断点标在第一步（打开橱柜）之前；因无视频、无状态日志，现象与根因留空。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "无视频、无状态日志、未交付轨迹（missing_trajectory=1）；所有节点 unknown、现象与根因留空并写 gap_zh，断点 首步前，与 03 对“agent 侧未交付”的规定一致。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:53:43Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-freeze-fruit-i00-privileged--cxuoqcb/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-freeze-fruit-i00-privileged--cxuoqcb/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "5016b69cc5e68b60f052a423c2dc24fe8dbea9bbb6f0971f4bfb62bc5240cde2",
+   "video_sha256": null,
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; no local video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/installing_a_fax_machine/trajectories/behavior-installing-a-fax-machine-i00-standard--tgrpgwy.json b/phaseR/behavior-1k/installing_a_fax_machine/trajectories/behavior-installing-a-fax-machine-i00-standard--tgrpgwy.json
new file mode 100644
index 0000000000000000000000000000000000000000..d3959f4887a52b93f8258f0d7e8bc5ebe8bc3b29
--- /dev/null
+++ b/phaseR/behavior-1k/installing_a_fax_machine/trajectories/behavior-installing-a-fax-machine-i00-standard--tgrpgwy.json
@@ -0,0 +1,184 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-installing-a-fax-machine-i00-standard--tgrpgwy",
+ "taskKey": "behavior-1k/installing_a_fax_machine",
+ "dag_sha256": "c1e9fa49e4b62a79bfa98de9cbe1d2a9e6e25dbc20b11272f9523f9bdc5e8754",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 2083,
+   "sim_steps": 2083,
+   "physics_steps": 8332,
+   "steps": 336,
+   "elapsed_s": 4603.6,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 0,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 1.0,
+   "goal_first_step": 2054,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-installing-a-fax-machine-i00-standard--tgrpgwy.mp4",
+  "duration_s": 69.5,
+  "fps": 10.0,
+  "frames": 695,
+  "sha256": "1ecb20d3ecb41c764d3f788849595912b655c76ee4662deaf32ce9f9e92cdbc5"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 2,
+  "final": 2,
+  "nodes": {
+   "fax_on_carrel": {
+    "method": "visual",
+    "intervals": [
+     [
+      65.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 65.0,
+      "log_step": null,
+      "text_zh": "观察：视频 64 s 腕部视角仍见夹爪夹持机身贴近白色桌面，65 s 夹爪松开离开机身，66 s 起机身平放于白色桌面、数字键盘朝上，头部视角可见传真机在隔间桌上；评分器 q_score=1.0、goal_first_step 2054（约 68.5 s）。推断：放置约在视频 65 s 达成，具体隔间桌编号未核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：夹爪夹住传真机机身搬到隔间白色桌面后松开（视频约 56-65 s）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "推断：轮式底盘驱动并承载夹持的传真机到隔间桌"
+     }
+    ]
+   },
+   "fax_switched_on": {
+    "method": "visual",
+    "intervals": [
+     [
+      68.7,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 68.7,
+      "log_step": null,
+      "text_zh": "观察：视频 66-68 s 腕部相机里夹爪手指压在红色圆形开关标记上，68.5 s 红色开始变色，69 s 起变为绿色并保持到视频末 69.4 s。推断：开关约在视频 68.7 s 翻开；评分器 goal_first_step 2054（2054/30≈68.5 s）与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "压/擦",
+      "note_zh": "观察：夹爪指尖压在红色圆形开关标记上（视频 66-69 s），标记变绿"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": null,
+ "summary_zh": "成功：标准模式单次实机回合，视频约 65 s 传真机放到隔间白色桌面并松手，约 68.7 s 手指按下红色开关标记变绿，两个节点均达成（评分器 q_score=1.0，goal_first_step 2054）。无状态日志，时间来自视频帧；隔间桌编号与前段搬运过程未逐帧核对。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "成功运行，两节点时间与帧一致（65 s 放到白桌松手，约68.5-69 s 红色开关变绿），评分器 q_score=1、goal_first_step 2054 与之吻合，无失败标注。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:53:56Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-installing-a-fax-machine-i00-standard--tgrpgwy/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-installing-a-fax-machine-i00-standard--tgrpgwy/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "418168cf97ce81db76052aeae14d7c68ee490c44d0ab910f0f2fa7f30ef8da2f",
+   "video_sha256": "1ecb20d3ecb41c764d3f788849595912b655c76ee4662deaf32ce9f9e92cdbc5",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/laying_tile_floors/trajectories/behavior-laying-tile-floors-i00-privileged--uyeat2h.json b/phaseR/behavior-1k/laying_tile_floors/trajectories/behavior-laying-tile-floors-i00-privileged--uyeat2h.json
new file mode 100644
index 0000000000000000000000000000000000000000..885d82da8a77bc8e02d8b7fb5fe0abf46f7d5eae
--- /dev/null
+++ b/phaseR/behavior-1k/laying_tile_floors/trajectories/behavior-laying-tile-floors-i00-privileged--uyeat2h.json
@@ -0,0 +1,239 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-laying-tile-floors-i00-privileged--uyeat2h",
+ "taskKey": "behavior-1k/laying_tile_floors",
+ "dag_sha256": "9bf4d7ab26900c399e507d461862a0b11aed154e1aacec0f08de6cb2afc3792c",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 509,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 659,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 977.3,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 18,
+   "agent_sim_steps": 0,
+   "agent_records": 4,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.825,
+   "peak_memory_gb": 12.892,
+   "peak_processes": 47,
+   "peak_disk_gb": 0.714,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-laying-tile-floors-i00-privileged--uyeat2h.mp4",
+  "duration_s": 21.967,
+  "fps": 30.0,
+  "frames": 659,
+  "sha256": "363ecae10e1d10924a45e7a3ad84321e8a995d3dcec5cd079509529a3c06be6c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 5,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bathroom_door_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 21.9,
+      "log_step": null,
+      "text_zh": "观察：视频 0-22 s 未见瓷砖被放到浴室地面；视频 21.9 s 画面朝向天花板与墙角。评分器 q_score=0、goal=0。未见带卫生间标识的门被推开形成通口（视频 5-6 s 仅见一扇木门边缘与会议室，无法确认为浴室门）。"
+     }
+    ]
+   },
+   "tiles_on_bathroom_floor_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 21.9,
+      "log_step": null,
+      "text_zh": "观察：视频 ≈4 s 与 11.5–13 s 头部相机在走廊黑色地面上见白色瓷砖，未被抓起或移动；视频 21.9 s 画面朝向天花板与墙角。评分器 q_score=0、goal=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "未知：画面未显示任何手指/夹爪与瓷砖接触（agent 自述 log step 108 称有尝试，属会话内证据，未核实，不作为接触）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：底盘沿走廊移动、进入会议室后折回至走廊瓷砖附近，未到浴室"
+     }
+    ]
+   },
+   "tiles_on_bathroom_floor_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 21.9,
+      "log_step": null,
+      "text_zh": "观察：视频 ≈4 s 与 11.5–13 s 头部相机在走廊黑色地面上见白色瓷砖，未被抓起或移动；视频 21.9 s 画面朝向天花板与墙角。评分器 q_score=0、goal=0。"
+     }
+    ]
+   },
+   "tiles_on_bathroom_floor_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 21.9,
+      "log_step": null,
+      "text_zh": "观察：视频 ≈4 s 与 11.5–13 s 头部相机在走廊黑色地面上见白色瓷砖，未被抓起或移动；视频 21.9 s 画面朝向天花板与墙角。评分器 q_score=0、goal=0。"
+     }
+    ]
+   },
+   "tiles_on_bathroom_floor_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 21.9,
+      "log_step": null,
+      "text_zh": "观察：视频 ≈4 s 与 11.5–13 s 头部相机在走廊黑色地面上见白色瓷砖，未被抓起或移动；视频 21.9 s 画面朝向天花板与墙角。评分器 q_score=0、goal=0。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "tiles_on_bathroom_floor_1",
+   "where": "操作中",
+   "t": 12.0,
+   "log_step": 108,
+   "evidence_zh": "观察：视频 5–10 s 驶入会议室后折回走廊；≈11.5–13 s 到达走廊瓷砖附近，瓷砖未被接触或移动；13 s 起头部相机滚转朝向天花板并静止至 21.9 s。门节点（bathroom_door_opened）及第1块砖从未被实际尝试放置。agent 自述（log step 108）称有失败抓取，未核实。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "观察：视频 ≈4 s 与 11.5–13 s 头部相机见走廊地面瓷砖，未被抓起或移动；13 s 起相机朝向天花板并静止至 21.9 s。视频未显示任何手指/夹爪与瓷砖接触，失败原因无法由被评分回放判定（agent 自述的抓取几何问题属会话测试，未核实，不记为根因）。无状态日志。",
+  "unmet": {
+   "tiles_on_bathroom_floor_1": false,
+   "tiles_on_bathroom_floor_2": false,
+   "tiles_on_bathroom_floor_3": false,
+   "tiles_on_bathroom_floor_4": false
+  }
+ },
+ "summary_zh": "观察：被评分的视频（22 s）中机器人先驶入会议室、折回走廊瓷砖附近，瓷砖未被接触或移动，13 s 起相机朝向天花板并静止，0/5 节点达成（q_score=0）。断点在第1块砖铺放（未开始操作）；根因无法由回放判定，仅有 agent 自述未核实。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点值（0/5）与 final 正确，但文件称“视频未拍到瓷砖”不实：视频 ≈4 s 与 11.5–13 s 在走廊地面清楚可见未被移动的瓷砖；contact_actual 与根因仅来自 agent 自述（会话测试，18 次 reset），不属于被评分回放的观察；视频实际所见（先驶入会议室、回到走廊瓷砖旁、13 s 起相机滚转朝向天花板并静止）未写入。",
+    "n_issues": 4,
+    "n_major": 3
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "第1轮问题已修正：节点0/5与q_score=0一致，视频所见（驶入会议室、折回走廊瓷砖旁、13 s起相机朝天花板静止）描述属实；无手指接触证据故不给现象/根因并写gap_zh，合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:39Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-laying-tile-floors-i00-privileged--uyeat2h/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-laying-tile-floors-i00-privileged--uyeat2h/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-laying-tile-floors-i00-privileged--uyeat2h/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-laying-tile-floors-i00-privileged--uyeat2h/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "0ad0eedf253e243aec6ba8e57c3b772d780f520102e8860d716a2b371d9aeee5",
+   "video_sha256": "363ecae10e1d10924a45e7a3ad84321e8a995d3dcec5cd079509529a3c06be6c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/loading_the_car/trajectories/behavior-loading-the-car-i00-privileged--zha4v28.json b/phaseR/behavior-1k/loading_the_car/trajectories/behavior-loading-the-car-i00-privileged--zha4v28.json
new file mode 100644
index 0000000000000000000000000000000000000000..3f82e1be8252f595aac1e006c5b41a8e0258ecc7
--- /dev/null
+++ b/phaseR/behavior-1k/loading_the_car/trajectories/behavior-loading-the-car-i00-privileged--zha4v28.json
@@ -0,0 +1,236 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-loading-the-car-i00-privileged--zha4v28",
+ "taskKey": "behavior-1k/loading_the_car",
+ "dag_sha256": "384f89102dcf627f89b410aa1d68d11d2f454502523ee32a707397692405b580",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 1009,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 0,
+   "goal_first_step": -1,
+   "sim_steps": 356,
+   "fallen": 1,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 590.3,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 2,
+   "agent_sim_steps": 0,
+   "agent_records": 1,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-loading-the-car-i00-privileged--zha4v28.mp4",
+  "duration_s": 11.867,
+  "fps": 30.0,
+  "frames": 356,
+  "sha256": "b2f7137afdf8db00aad673c3cb18d207e97e63dcdf103390e3b459583dbc0be8"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 4,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "car_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-6 s 机器人在室内转场，画面倾斜翻转，约视频 6.5–7 s 起头部相机贴近木地板、此后至 11.9 s 保持不变（机器人已倒地）；全程未见汽车、收纳箱、相机或球拍。评分器字段 fallen=1、q_score=0。 汽车装载盖始终未见打开。"
+     }
+    ]
+   },
+   "camera_in_container": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-6 s 机器人在室内转场，画面倾斜翻转，约视频 6.5–7 s 起头部相机贴近木地板、此后至 11.9 s 保持不变（机器人已倒地）；全程未见汽车、收纳箱、相机或球拍。评分器字段 fallen=1、q_score=0。"
+     }
+    ]
+   },
+   "container_in_car": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-6 s 机器人在室内转场，画面倾斜翻转，约视频 6.5–7 s 起头部相机贴近木地板、此后至 11.9 s 保持不变（机器人已倒地）；全程未见汽车、收纳箱、相机或球拍。评分器字段 fallen=1、q_score=0。"
+     }
+    ]
+   },
+   "racket_in_car": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-6 s 机器人在室内转场，画面倾斜翻转，约视频 6.5–7 s 起头部相机贴近木地板、此后至 11.9 s 保持不变（机器人已倒地）；全程未见汽车、收纳箱、相机或球拍。评分器字段 fallen=1、q_score=0。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "无量纲",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "car_opened",
+   "t": 6.5,
+   "log_step": 103,
+   "where": "转场",
+   "evidence_zh": "观察：视频约 0-6.5 s 机器人底盘在客厅驱动途中倾倒，之后一直躺在地板上，从未到达车库汽车旁；评分器 fallen=1。agent 最终自述“the robot tipped over during the living-room transit”（log step 103）。推断：agent 的计划先开汽车（最后报告称汽车盖是核心障碍），故断点取 打开汽车。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 6.5,
+    "log_step": 103,
+    "evidence_zh": "观察：视频约 6.5–7 s 起头部相机贴地、画面不再变化，直到 11.9 s；评分器 fallen=1。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "alt_feasible": "未知",
+    "evidence_zh": "agent 自述底盘速度 0.12/0.30 导致倾倒，后改为 0.05/0.10（log step 103）；交付的轨迹仍是这次倾倒的失败运行。子类型不明，停在控制。",
+    "source_zh": "agent 自述"
+   }
+  ],
+  "gap_zh": "无状态日志，视频为头/腕相机视角，看不到机器人整体，倒地精确时刻只能估计（约视频 6.5–7 s）；断点节点依 agent 计划推断。",
+  "unmet": {
+   "camera_in_container": false,
+   "container_in_car": false,
+   "racket_in_car": false
+  }
+ },
+ "summary_zh": "视频 11.9 s 内机器人在室内转场时倾倒并躺在地上，四个节点（开车、相机入箱、箱入车、球拍入车）均未达成，q_score=0、fallen=1。断点为打开汽车（转场），现象全身·平衡，根因仅有 agent 自述的底盘速度过高，归控制。主要缺口：无状态日志，倒地时刻与断点节点靠视频和 agent 自述推断。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "断点（转场、倾倒）与现象大体成立，但四个节点和断点处的观察“全程未见汽车”与视频不符：约 5 s 帧中可见红色汽车。倒地时刻也应在约 6.5–7 s，而非 6 s。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "复查视频 3–7 s：无红色汽车，仅室内天花板、白墙、木门，6.5–7 s 起头相机贴地，与标签一致；第1轮 I1 无证据，不重复。四节点未达成、转场倾倒断点、全身·平衡与控制根因均成立。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:34Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-loading-the-car-i00-privileged--zha4v28/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-loading-the-car-i00-privileged--zha4v28/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-loading-the-car-i00-privileged--zha4v28/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-loading-the-car-i00-privileged--zha4v28/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "2c0faa80d7fee9df594fda8c20ca07c3ab0172dd67dadf563c43e627e0e4f4e2",
+   "video_sha256": "b2f7137afdf8db00aad673c3cb18d207e97e63dcdf103390e3b459583dbc0be8",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/loading_the_car/trajectories/behavior-loading-the-car-i00-standard--axaoubt.json b/phaseR/behavior-1k/loading_the_car/trajectories/behavior-loading-the-car-i00-standard--axaoubt.json
new file mode 100644
index 0000000000000000000000000000000000000000..5d577022369d9a4c7728b0e4318dd0f9cd94fc54
--- /dev/null
+++ b/phaseR/behavior-1k/loading_the_car/trajectories/behavior-loading-the-car-i00-standard--axaoubt.json
@@ -0,0 +1,239 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-loading-the-car-i00-standard--axaoubt",
+ "taskKey": "behavior-1k/loading_the_car",
+ "dag_sha256": "384f89102dcf627f89b410aa1d68d11d2f454502523ee32a707397692405b580",
+ "outcome": {
+  "success": 0,
+  "reward": 0.3333333333333333,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 38516,
+   "sim_steps": 38516,
+   "physics_steps": 154064,
+   "steps": 38516,
+   "elapsed_s": 13407.2,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "q_score": 0.3333333333333333,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-loading-the-car-i00-standard--axaoubt/replay.mp4",
+  "duration_s": 1283.867,
+  "fps": 30.0,
+  "frames": 38516,
+  "sha256": "fa708ea2706aca0fa9187394d40eaf929f702c5cff38cc14a62454976d378e85"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 4,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "car_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1100,
+      "log_step": null,
+      "text_zh": "观察：视频 约 1000-1280 s 的头部相机多帧显示机器人在车库里围着红色轿车，夹爪在车门把手/车尾盖边缘附近；所有帧里车门和后备箱均保持合拢，未见开口。推断：打开汽车未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：右/左夹爪多次伸向车门把手、后备箱边缘和后视镜，抓握均未带动门盖。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在车周围多次移动以对准把手/后备箱。"
+     }
+    ]
+   },
+   "camera_in_container": {
+    "method": "visual",
+    "intervals": [
+     [
+      324,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 324,
+      "log_step": 229,
+      "text_zh": "观察：视频 约 324 s 左腕相机/头部画面显示数码相机平放在蓝色收纳箱内，随后搬运和放在车库地面时（视频 约 410 s 起）箱内仍可见黑色相机。视频 约 232-260 s 相机曾掉到箱旁地面（agent 自述 log step 229），此前一段是否算在箱内未能精确定位；q_score=1/3 与只有此关系成立一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右夹爪夹起相机并放入收纳箱；中途相机滑落后重新抓取再放入。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在桌与箱之间移动。"
+     }
+    ]
+   },
+   "container_in_car": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1200,
+      "log_step": null,
+      "text_zh": "观察：收纳箱自视频 约 450 s 起一直放在车库地面、车旁，从未进入车内（车门未开）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "夹爪夹住箱沿搬起收纳箱。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘载箱从客厅经走廊开到车库，放在车库地面。"
+     }
+    ]
+   },
+   "racket_in_car": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1200,
+      "log_step": 265,
+      "text_zh": "观察：球拍曾被左夹爪抓起，之后（agent 自述 log step 265）被放回桌上，视频后段未见进车。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "左夹爪抓握球拍柄抬起，后放回桌面。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "car_opened",
+   "t": 1000,
+   "log_step": 367,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 约 450 s 起机器人把装着相机的收纳箱放在车库地面，其后到视频结束（1283.9 s）一直在车周围用夹爪试探门把手、后备箱和后视镜，车始终未打开，箱和球拍都没进车。log step 367-785 记录了多次失败的把手抓取。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 1000,
+    "log_step": 386,
+    "evidence_zh": "观察：视频 约 1000-1280 s 夹爪靠近把手/车尾盖但车门始终没有移动；agent 自述多次“抓握闭合为空”（log step 531、745）。推断：需要的抓握接触从未建立。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "evidence_zh": "agent 自述手指被门板/把手凹槽挡住，试了把手、后备箱边缘、后视镜、门上沿等多种抓法均失败（log step 386、548、745）；是否有可行抓法未验证。",
+    "source_zh": "agent 自述"
+   }
+  ],
+  "gap_zh": "视频只有头部/腕部视角，把手处接触细节有遮挡；相机第一次放入箱内的确切时刻（约视频 200-230 s）未能精确定位。视频 1284 s 对应 38516 控制步，仅覆盖实际操作，agent 的启动等待不在视频内。",
+  "unmet": {
+   "camera_in_container": true,
+   "container_in_car": false,
+   "racket_in_car": false
+  }
+ },
+ "summary_zh": "观察：视频约 324 s 相机重新放入收纳箱并保持到结束（视频 1283.9 s），收纳箱搬到车库地面，但汽车始终未打开，容器和球拍均未入车，4 个节点中仅相机入箱成立（与 q_score 1/3 一致）。断点是打开汽车：夹爪反复抓不住车门把手、后备箱边缘和后视镜。证据缺口：把手处接触被遮挡，相机首次入箱时间不精确。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "抽查视频：约250 s相机落在箱旁地面、约270 s重新被抓起、324 s后入箱，与标注一致；1100/1280 s车库帧车门/后备箱始终合拢，容器在车外，与 q=1/3 一致。断点 car_opened、没建立接触、抓取和接触设计（alt_feasible 未知）可接受。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:55:34Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-loading-the-car-i00-standard--axaoubt/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-loading-the-car-i00-standard--axaoubt/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "c9bbfce43cce867ef636c346f7550bead99bcb96faf714cc731181bb935d0215",
+   "video_sha256": "fa708ea2706aca0fa9187394d40eaf929f702c5cff38cc14a62454976d378e85",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/make_pizza/trajectories/behavior-make-pizza-i00-privileged--qj3nshn.json b/phaseR/behavior-1k/make_pizza/trajectories/behavior-make-pizza-i00-privileged--qj3nshn.json
new file mode 100644
index 0000000000000000000000000000000000000000..3f85f7468301c78ea4f39ff74af0a8c2a65a1c22
--- /dev/null
+++ b/phaseR/behavior-1k/make_pizza/trajectories/behavior-make-pizza-i00-privileged--qj3nshn.json
@@ -0,0 +1,407 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-make-pizza-i00-privileged--qj3nshn",
+ "taskKey": "behavior-1k/make_pizza",
+ "dag_sha256": "d6d9df11b8f00b7e0bec05341901a476ef340b5195498408f6ce96ab9b7f72cc",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 60,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 210,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 609.4,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 1,
+   "agent_sim_steps": 0,
+   "agent_records": 1,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-make-pizza-i00-privileged--qj3nshn.mp4",
+  "duration_s": 7.0,
+  "fps": 30.0,
+  "frames": 210,
+  "sha256": "1532182d7f79b886043dd51d90a881f1de598420f510e51ceacb9e837086840b"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 19,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "oven_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "oven_on": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "cheese_on_dough": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "pepperoni_on_dough_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "pepperoni_on_dough_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "pepperoni_on_dough_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "pepperoni_on_dough_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "mushrooms_halved_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "mushrooms_halved_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "mushroom_halves_on_dough_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "mushroom_halves_on_dough_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "mushroom_halves_on_dough_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "mushroom_halves_on_dough_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "onion_diced": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "diced_onion_on_dough": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "dough_in_oven": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "pizza_baked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   },
+   "pizza_on_sheet": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.9,
+      "log_step": 99,
+      "text_zh": "观察：视频 0-7.0 s 机器人静止不动，固定餐厅视角，冰箱/烤箱/台面均未被操作；log step 99 交付的轨迹仅为 reset 后 hold 90 步；评分 q_score=0。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "比例",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "fridge_opened",
+   "t": 0,
+   "log_step": 99,
+   "where": "首步前",
+   "evidence_zh": "观察：视频 0-7.0 s 机器人完全不动；log step 99 交付的是 reset 后空等（hold）的最小轨迹，q_score 0，第一步（打开冰箱）从未开始。"
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0,
+    "log_step": 99,
+    "evidence_zh": "观察：整段视频机器人不动，无任何接触；交付轨迹仅 hold。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "交付的是 reset 后 hold 的最小轨迹（log step 99），整段 0-7.0 s 视频无动作；agent 在 session 中探测烤箱门抓取（log 85-91）后预算耗尽，只来得及保存空等轨迹。无证据支持规划或控制类根因，故根因为空。",
+  "unmet": {
+   "pizza_baked": false,
+   "pizza_on_sheet": false
+  }
+ },
+ "summary_zh": "观察：视频 0-7.0 s 机器人静止，交付轨迹只是 reset 后空等（log step 99），19 个节点均未达成，q_score 0。断在第一步打开冰箱之前（没尝试）；agent 预算耗尽后只保存了空等轨迹，根因证据不足。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 8,
+    "summary_zh": "节点全 false、断点 fridge_opened/首步前/没尝试、根因为空均与证据一致（交付为 reset 后 hold，q_score=0，类同 FAIL-22）。仅 gap_zh 与 summary 把时间“耗在离线 IK 检查上”的说法缺乏证据，属 FAIL-4 的时间分配解读。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "19 个节点全 false、断点 fridge_opened/首步前/没尝试、根因为空均与证据一致；第 1 轮的 gap_zh 问题已修正。validator 通过。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:22Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-make-pizza-i00-privileged--qj3nshn/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-make-pizza-i00-privileged--qj3nshn/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-make-pizza-i00-privileged--qj3nshn/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-make-pizza-i00-privileged--qj3nshn/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6944980d49512188d34251057d511fac6aba387e9a0f0e09d5731bc7ac466076",
+   "video_sha256": "1532182d7f79b886043dd51d90a881f1de598420f510e51ceacb9e837086840b",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/organizing_art_supplies/trajectories/behavior-organizing-art-supplies-i00-privileged--dkrxdre.json b/phaseR/behavior-1k/organizing_art_supplies/trajectories/behavior-organizing-art-supplies-i00-privileged--dkrxdre.json
new file mode 100644
index 0000000000000000000000000000000000000000..00ce0337478efe7c5304b603a94eacc17c69f757
--- /dev/null
+++ b/phaseR/behavior-1k/organizing_art_supplies/trajectories/behavior-organizing-art-supplies-i00-privileged--dkrxdre.json
@@ -0,0 +1,256 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-organizing-art-supplies-i00-privileged--dkrxdre",
+ "taskKey": "behavior-1k/organizing_art_supplies",
+ "dag_sha256": "a05b0a960700fb93d0f7eacd83754f0d4bb6000a0831d5e2a7b21bdcc373cb9d",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "deterministic": 0,
+   "missing_trajectory": 1,
+   "grader_error": 0,
+   "infra_error": 0,
+   "agent_sessions": 1,
+   "agent_resets": 5,
+   "agent_sim_steps": 0,
+   "agent_records": 2,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.303,
+   "peak_memory_gb": 13.253,
+   "peak_processes": 45,
+   "peak_disk_gb": 0.693,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "none",
+  "url": null,
+  "duration_s": null,
+  "fps": null,
+  "frames": null,
+  "sha256": null
+ },
+ "state": null,
+ "progress": {
+  "method": "none",
+  "denominator": 5,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "glue_in_bag": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 107,
+      "text_zh": "观察：无视频、无状态日志；run.json 的 reward 中 missing_trajectory=1（没有交付轨迹），agent 在 log step 107 仍在其会话内轮询第 3 次试跑，未保存任何轨迹。无法判定该节点的视频区间；评分器仅给出 success=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，没有被评分的尝试。"
+     }
+    ]
+   },
+   "eraser_in_bag": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 107,
+      "text_zh": "观察：无视频、无状态日志；run.json 的 reward 中 missing_trajectory=1（没有交付轨迹），agent 在 log step 107 仍在其会话内轮询第 3 次试跑，未保存任何轨迹。无法判定该节点的视频区间；评分器仅给出 success=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，没有被评分的尝试。"
+     }
+    ]
+   },
+   "brush_in_bag": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 107,
+      "text_zh": "观察：无视频、无状态日志；run.json 的 reward 中 missing_trajectory=1（没有交付轨迹），agent 在 log step 107 仍在其会话内轮询第 3 次试跑，未保存任何轨迹。无法判定该节点的视频区间；评分器仅给出 success=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，没有被评分的尝试。"
+     }
+    ]
+   },
+   "marker_in_bag": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 107,
+      "text_zh": "观察：无视频、无状态日志；run.json 的 reward 中 missing_trajectory=1（没有交付轨迹），agent 在 log step 107 仍在其会话内轮询第 3 次试跑，未保存任何轨迹。无法判定该节点的视频区间；评分器仅给出 success=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，没有被评分的尝试。"
+     }
+    ]
+   },
+   "bag_on_desk": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 107,
+      "text_zh": "观察：无视频、无状态日志；run.json 的 reward 中 missing_trajectory=1（没有交付轨迹），agent 在 log step 107 仍在其会话内轮询第 3 次试跑，未保存任何轨迹。无法判定该节点的视频区间；评分器仅给出 success=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，没有被评分的尝试。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": null,
+   "unit": "1",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "glue_in_bag",
+   "t": null,
+   "log_step": 107,
+   "where": "首步前",
+   "evidence_zh": "观察：reward.missing_trajectory=1、agent_replays=0；log step 89 记录 run1 失败并保存了一条轨迹，log step 90 将其删除；log step 98-107 为后续试跑，轨迹也未保存（agent 工作约 7158 s 时仍在轮询试跑，总用时 7202 s）。推断：评分的尝试中没有任何交付的动作，第一步（把胶棒放进袋子）从未被评分。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "没有视频、没有状态日志，且 agent 没有交付轨迹（missing_trajectory=1），故无法观察被评分的尝试。会话内多次试跑（拾取细物失败、提袋滑脱）只是会话证据，不能作为现象；超时本身不是现象或根因。按规则交付缺失、原因在 agent 一侧时两层留空。",
+  "unmet": {
+   "glue_in_bag": null,
+   "eraser_in_bag": null,
+   "brush_in_bag": null,
+   "marker_in_bag": null,
+   "bag_on_desk": null
+  }
+ },
+ "summary_zh": "观察：agent 用了约 7200 s 反复在模拟器里试跑（log step 44-107），拾取细小物品（画笔、马克笔）和提起托特袋都不稳定，最终没有保存任何轨迹（missing_trajectory=1）。被评分的尝试为空，5 个节点均无法由视频判定（无视频、无状态日志），断点记为 glue_in_bag·首步前；现象与根因因无证据留空。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "无视频、无状态日志、交付缺失（missing_trajectory=1），节点全部 unknown、现象与根因留空并写 gap_zh，符合 03 的 agent 侧无交付规则。仅有一处 log 步号引用不准。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "无视频、无状态日志、无交付轨迹（missing_trajectory=1），节点全部 unknown、现象与根因留空并写 gap_zh，符合 03 对 agent 侧无交付的规则；上一轮的步号问题已修正。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:25Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-organizing-art-supplies-i00-privileged--dkrxdre/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-organizing-art-supplies-i00-privileged--dkrxdre/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-organizing-art-supplies-i00-privileged--dkrxdre/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-organizing-art-supplies-i00-privileged--dkrxdre/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "e422bd03c043c36e5d3d5885bb64db5309562acbc6c0933c1884059c4cda8f85",
+   "video_sha256": null,
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; no local video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/organizing_art_supplies/trajectories/behavior-organizing-art-supplies-i00-standard--4zorxtn.json b/phaseR/behavior-1k/organizing_art_supplies/trajectories/behavior-organizing-art-supplies-i00-standard--4zorxtn.json
new file mode 100644
index 0000000000000000000000000000000000000000..393576ce406febd282b2a04ade5e9f86dd08526e
--- /dev/null
+++ b/phaseR/behavior-1k/organizing_art_supplies/trajectories/behavior-organizing-art-supplies-i00-standard--4zorxtn.json
@@ -0,0 +1,279 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-organizing-art-supplies-i00-standard--4zorxtn",
+ "taskKey": "behavior-1k/organizing_art_supplies",
+ "dag_sha256": "a05b0a960700fb93d0f7eacd83754f0d4bb6000a0831d5e2a7b21bdcc373cb9d",
+ "outcome": {
+  "success": 0,
+  "reward": 0.8,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 40047,
+   "sim_steps": 40047,
+   "physics_steps": 160188,
+   "steps": 13153,
+   "elapsed_s": 9932.5,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "replay_error": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "q_score": 0.8,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.455,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.388,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-organizing-art-supplies-i00-standard--4zorxtn/replay.mp4",
+  "duration_s": 1334.9,
+  "fps": 30.0,
+  "frames": 40047,
+  "sha256": "16c706292003160d0838fd457a583c050c72d604e9b5c90e83b4ef1c09d7f918"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 5,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "glue_in_bag": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 1250,
+      "log_step": null,
+      "text_zh": "观察：视频 1250 s 与 1334 s 书桌上仍可见散落的小件文具（绿色胶棒样与红/深色笔），袋口被遮挡，无法逐件确认哪件已入袋；评分器 q_score 0.8（约 4/5 个目标）与此不完全一致，推断：无状态日志，逐件入袋无法判定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "帧中无法确认该件文具是否被夹爪抓握放入。"
+     }
+    ]
+   },
+   "eraser_in_bag": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 1250,
+      "log_step": null,
+      "text_zh": "观察：视频 1250 s 与 1334 s 书桌上仍可见散落的小件文具（绿色胶棒样与红/深色笔），袋口被遮挡，无法逐件确认哪件已入袋；评分器 q_score 0.8（约 4/5 个目标）与此不完全一致，推断：无状态日志，逐件入袋无法判定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "帧中无法确认该件文具是否被夹爪抓握放入。"
+     }
+    ]
+   },
+   "brush_in_bag": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 1250,
+      "log_step": null,
+      "text_zh": "观察：视频 1250 s 与 1334 s 书桌上仍可见散落的小件文具（绿色胶棒样与红/深色笔），袋口被遮挡，无法逐件确认哪件已入袋；评分器 q_score 0.8（约 4/5 个目标）与此不完全一致，推断：无状态日志，逐件入袋无法判定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "帧中无法确认该件文具是否被夹爪抓握放入。"
+     }
+    ]
+   },
+   "marker_in_bag": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 1250,
+      "log_step": null,
+      "text_zh": "观察：视频 1250 s 与 1334 s 书桌上仍可见散落的小件文具（绿色胶棒样与红/深色笔），袋口被遮挡，无法逐件确认哪件已入袋；评分器 q_score 0.8（约 4/5 个目标）与此不完全一致，推断：无状态日志，逐件入袋无法判定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "帧中无法确认该件文具是否被夹爪抓握放入。"
+     }
+    ]
+   },
+   "bag_on_desk": {
+    "method": "visual",
+    "intervals": [
+     [
+      60,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 60,
+      "log_step": null,
+      "text_zh": "观察：视频 60 s 起头部相机可见手提袋在书桌桌面（夹爪在袋口），此后 90–260 s 袋子多次被夹爪提起/翻动后又放回桌面，260 s 起至 1334 s 帧袋子均在桌面（1312 s、1334 s 袋在桌面、夹爪在旁，靠窗角落）。130–250 s 间短暂提起的间隙因稀疏抽帧无法精确切分，按连续区间记，误差约 ±30 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：夹爪抓握/扶持袋提手，反复提起并放回桌面。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.8,
+   "unit": "1",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "unknown",
+   "t": null,
+   "log_step": null,
+   "where": "未知",
+   "evidence_zh": "推断：q_score 0.8 表示约一个目标不成立，末帧袋在桌面，但画面无法确定哪件文具未入袋，断点无法定位。agent 的 token 预算在 log step 529 用尽，预算耗尽不作现象。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无状态日志，仅稀疏抽帧：q_score 0.8 推断恰有一个目标不成立（推断，非观察），末帧袋子在桌面，故更可能是某件文具未入袋，但画面无法确定是哪件，断点无法定位，现象与根因留空。",
+  "unmet": {
+   "glue_in_bag": null,
+   "eraser_in_bag": null,
+   "brush_in_bag": null,
+   "marker_in_bag": null,
+   "bag_on_desk": true
+  }
+ },
+ "summary_zh": "观察：视频 1334.9 s，机器人把手提袋放在书桌上（约 60 s 起，其间多次被提起又放回），末帧袋子仍在桌面；桌面仍可见散落文具。评分 q_score 0.8（推断：约 4/5 目标成立），success 0。四件文具逐件入袋因袋口遮挡和无状态日志无法判定，断点无法定位。agent 在 log step 529 因 token 预算用尽结束。",
+ "conf": 3,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "bag_on_desk 的时间和终值与画面不符：袋子最迟约 60 s 起就在桌面（并非 240 s），1334 s 末帧袋子仍在桌面上（靠窗角落），并未被带离书桌；q_score 0.8 说明五个目标中仅一个不成立，被标 false 的 bag_on_desk 与末帧矛盾，断点因此建在无依据的节点上。",
+    "n_issues": 3,
+    "n_major": 3
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "第一轮问题已修正：bag_on_desk 终值 true，与 1334 s 末帧袋在桌面一致；四件文具无状态日志、袋口遮挡，q_score 0.8 只能推断一个目标不成立，故节点 unknown、断点 null 并写 gap_zh，合理。未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:06Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-organizing-art-supplies-i00-standard--4zorxtn/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-organizing-art-supplies-i00-standard--4zorxtn/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-organizing-art-supplies-i00-standard--4zorxtn/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-organizing-art-supplies-i00-standard--4zorxtn/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "0d8f685990b3b0d93ec8f30213a3b7176cc8b03483edae4211ee05f04a72c16d",
+   "video_sha256": "16c706292003160d0838fd457a583c050c72d604e9b5c90e83b4ef1c09d7f918",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/outfit_a_basic_toolbox/trajectories/behavior-outfit-a-basic-toolbox-i00-standard--cmehrpb.json b/phaseR/behavior-1k/outfit_a_basic_toolbox/trajectories/behavior-outfit-a-basic-toolbox-i00-standard--cmehrpb.json
new file mode 100644
index 0000000000000000000000000000000000000000..ad84f0ead9e09eea0cced55c5671b9277564e825
--- /dev/null
+++ b/phaseR/behavior-1k/outfit_a_basic_toolbox/trajectories/behavior-outfit-a-basic-toolbox-i00-standard--cmehrpb.json
@@ -0,0 +1,375 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-outfit-a-basic-toolbox-i00-standard--cmehrpb",
+ "taskKey": "behavior-1k/outfit_a_basic_toolbox",
+ "dag_sha256": "4139a9fc7f1a291c72fd33a94dba792038f26e761f4cdee96373259f36e5a8ad",
+ "outcome": {
+  "success": 0,
+  "reward": 0.7142857142857143,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "n_actions": 24374,
+   "sim_steps": 24374,
+   "physics_steps": 97496,
+   "steps": 23813,
+   "elapsed_s": 11237.6,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "q_score": 0.7142857142857143,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-outfit-a-basic-toolbox-i00-standard--cmehrpb/replay.mp4",
+  "duration_s": 812.467,
+  "fps": 30.0,
+  "frames": 24374,
+  "sha256": "08732a220a526ee327e8322ff0845512b68fd819a2aed69790fabf3c61d1bf92"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 8,
+  "base": 2,
+  "peak": 7,
+  "final": 7,
+  "nodes": {
+   "toolbox_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      135,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 135,
+      "log_step": null,
+      "text_zh": "观察：视频约 135 s 头部相机可见机器人夹爪掀起箱盖，箱口露出（此前 100-130 s 箱盖仍闭合）。事件节点，之后关盖不撤销。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "机器人夹爪从盖缘掀开箱盖（帧上无法确认是夹住还是拨开）"
+     }
+    ]
+   },
+   "drill_in_toolbox": {
+    "method": "visual",
+    "intervals": [
+     [
+      555,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 555,
+      "log_step": null,
+      "text_zh": "观察：视频约 555 s 起电钻已在箱内（电钻 555 s 离开桌面进箱）。推断：评分器 q_score=5/7，与五件工具入箱、箱盖未关一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "机器人单臂夹爪夹住工具后送入箱内释放"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "轮式底盘支承双臂，作业中有移动"
+     }
+    ]
+   },
+   "pliers_in_toolbox": {
+    "method": "visual",
+    "intervals": [
+     [
+      600,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 600,
+      "log_step": null,
+      "text_zh": "观察：视频约 600 s 起钳子已在箱内（箱内工具逐件增多）；该工具入箱时刻只能由 5 s 间隔帧大致推断，误差约 ±15 s（黄色工具在帧中难区分）。推断：评分器 q_score=5/7，与五件工具入箱、箱盖未关一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "机器人单臂夹爪夹住工具后送入箱内释放"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "轮式底盘支承双臂，作业中有移动"
+     }
+    ]
+   },
+   "flashlight_in_toolbox": {
+    "method": "visual",
+    "intervals": [
+     [
+      630,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 630,
+      "log_step": null,
+      "text_zh": "观察：视频约 630 s 起手电筒已在箱内（箱内工具逐件增多）；该工具入箱时刻只能由 5 s 间隔帧大致推断，误差约 ±15 s（黄色工具在帧中难区分）。推断：评分器 q_score=5/7，与五件工具入箱、箱盖未关一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "机器人单臂夹爪夹住工具后送入箱内释放"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "轮式底盘支承双臂，作业中有移动"
+     }
+    ]
+   },
+   "allen_wrench_in_toolbox": {
+    "method": "visual",
+    "intervals": [
+     [
+      696,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 696,
+      "log_step": null,
+      "text_zh": "观察：视频约 696 s 起内六角扳手已在箱内（扳手 690 s 仍在桌面，696 s 已在箱内）。推断：评分器 q_score=5/7，与五件工具入箱、箱盖未关一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "机器人单臂夹爪夹住工具后送入箱内释放"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "轮式底盘支承双臂，作业中有移动"
+     }
+    ]
+   },
+   "screwdriver_in_toolbox": {
+    "method": "visual",
+    "intervals": [
+     [
+      660,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 660,
+      "log_step": null,
+      "text_zh": "观察：视频约 660 s 起螺丝刀已在箱内（箱内工具逐件增多）；该工具入箱时刻只能由 5 s 间隔帧大致推断，误差约 ±15 s（黄色工具在帧中难区分）。推断：评分器 q_score=5/7，与五件工具入箱、箱盖未关一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "机器人单臂夹爪夹住工具后送入箱内释放"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "轮式底盘支承双臂，作业中有移动"
+     }
+    ]
+   },
+   "toolbox_on_table": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：视频全程工具箱在桌面上（0 s 起，末帧仍在桌上，盖敞开）；约 420-540 s 箱子被推动/歪斜但未离桌。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "夹爪开合箱盖时推动箱体，箱体未离开桌面"
+     }
+    ]
+   },
+   "toolbox_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      135
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 812,
+      "log_step": null,
+      "text_zh": "观察：0-135 s 箱盖闭合；约 135 s 起打开装箱；约 400 s 曾见箱盖合上但工具未装完（此处箱盖状态仅见于 20 s 间隔帧，未并入区间）；末段 700-812 s 箱盖敞开约 90°，机器人夹爪在盖缘附近尝试合盖，末帧仍敞开。推断：q_score 5/7 未计入合盖。"
+     },
+     {
+      "t": 400,
+      "log_step": null,
+      "text_zh": "观察：视频 400 s 帧工具箱盖看似合着、手电筒仍在桌面；该次关闭不属于装完五件之后，不计入末态。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "机器人夹爪在盖缘尝试把盖推回合上，未成功"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "toolbox_closed",
+   "t": 750,
+   "log_step": 790,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 696 s 起五件工具均在箱内，700-812 s 箱盖敞开约 90°，机器人夹爪反复靠近盖缘，末帧（812 s）盖仍敞开。agent 自述“手柄离货架太近无法抓稳，改用盖上缘”（log step 790）。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无状态日志；合盖失败的具体接触现象无法由视频与评分器判定：视频 700-812 s 仅见盖敞开约 90°、夹爪在盖缘附近（log step 786-795 显示 agent 以夹爪触盖缘/手柄尝试，接触可能已建立但未带动盖），故现象与根因留空。钳子、手电筒、螺丝刀入箱时刻为 5 s 间隔帧推断，不够精确。",
+  "unmet": {
+   "drill_in_toolbox": true,
+   "pliers_in_toolbox": true,
+   "flashlight_in_toolbox": true,
+   "allen_wrench_in_toolbox": true,
+   "screwdriver_in_toolbox": true,
+   "toolbox_on_table": true,
+   "toolbox_closed": false
+  }
+ },
+ "summary_zh": "观察：五件工具约 555-696 s（视频）陆续入箱，工具箱始终在桌上，评分 q=5/7。推断：只差合盖——700-812 s 箱盖敞开，机器人夹爪靠近盖缘却合不上，断点为 关上工具箱（toolbox_closed）。主要缺口：无状态日志，三件工具的入箱时间与合盖失败的具体接触只能由帧推断。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "进度节点（5 件入箱、开箱 135 s、桌上、合盖未成）与评分 q=5/7 及抽帧一致，断点 toolbox_closed 合理；但唯一现象“没建立接触”仅由头部相机“看不到推动”推出，文件自己的 gap_zh 又承认夹爪是否碰到盖看不清，属把推断当观察，且 log 显示 agent 在 786-795 步持续用右臂夹盖缘/手柄尝试合盖，视频末帧夹爪就在盖缘。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "第 1 轮意见已按证据采纳：现象与根因留空并有 gap_zh；进度节点与 q=5/7 一致，断点 toolbox_closed 合理。未发现新错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:43Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-outfit-a-basic-toolbox-i00-standard--cmehrpb/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-outfit-a-basic-toolbox-i00-standard--cmehrpb/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-outfit-a-basic-toolbox-i00-standard--cmehrpb/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-outfit-a-basic-toolbox-i00-standard--cmehrpb/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "d3999c893bd935d33729f8915890a5a147408c950cea380c2c99bce99c40436b",
+   "video_sha256": "08732a220a526ee327e8322ff0845512b68fd819a2aed69790fabf3c61d1bf92",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/packing_meal_for_delivery/trajectories/behavior-packing-meal-for-delivery-i00-privileged--6usjiz3.json b/phaseR/behavior-1k/packing_meal_for_delivery/trajectories/behavior-packing-meal-for-delivery-i00-privileged--6usjiz3.json
new file mode 100644
index 0000000000000000000000000000000000000000..00122fe0e953d55e3f523c76a63658fd843af245
--- /dev/null
+++ b/phaseR/behavior-1k/packing_meal_for_delivery/trajectories/behavior-packing-meal-for-delivery-i00-privileged--6usjiz3.json
@@ -0,0 +1,239 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-packing-meal-for-delivery-i00-privileged--6usjiz3",
+ "taskKey": "behavior-1k/packing_meal_for_delivery",
+ "dag_sha256": "4fcaf2f7983d9623e89a1c1b8bf9493a93ef2e245dcdd37262fa201c9fa6cd68",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 0,
+   "n_actions": 1557,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 1707,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 780.0,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 9,
+   "agent_sim_steps": 0,
+   "agent_records": 4,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.165,
+   "peak_memory_gb": 12.208,
+   "peak_processes": 44,
+   "peak_disk_gb": 0.561,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-packing-meal-for-delivery-i00-privileged--6usjiz3.mp4",
+  "duration_s": 56.9,
+  "fps": 30.0,
+  "frames": 1707,
+  "sha256": "856be27b02e50711c51c5349b574edc5659933674d390654ccd2709879e5d87d"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 6,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hamburgers_in_sacks_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 41.0,
+      "log_step": null,
+      "text_zh": "观察：视频 9 s 可见一只纸袋，视频约 21-40 s 机器人夹爪抓着汉堡在厨房中搬运，视频 41-56 s 只见储物箱和袋子在台面上，未见汉堡落入任何纸袋；评分器 q_score=0、goal=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频约 21-40 s 机器人夹爪抓住包装汉堡并搬运；底盘驱动移动。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驱动携带汉堡穿过厨房。"
+     }
+    ]
+   },
+   "hamburgers_in_sacks_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 41.0,
+      "log_step": null,
+      "text_zh": "观察：全程未见袋内有汉堡、也未见袋被放到储物箱顶面；推断：评分器 q_score=0 与此一致。"
+     }
+    ]
+   },
+   "hamburgers_in_sacks_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 41.0,
+      "log_step": null,
+      "text_zh": "观察：全程未见袋内有汉堡、也未见袋被放到储物箱顶面；推断：评分器 q_score=0 与此一致。"
+     }
+    ]
+   },
+   "sacks_on_box_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 41.0,
+      "log_step": null,
+      "text_zh": "观察：全程未见袋内有汉堡、也未见袋被放到储物箱顶面；推断：评分器 q_score=0 与此一致。"
+     }
+    ]
+   },
+   "sacks_on_box_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 41.0,
+      "log_step": null,
+      "text_zh": "观察：全程未见袋内有汉堡、也未见袋被放到储物箱顶面；推断：评分器 q_score=0 与此一致。"
+     }
+    ]
+   },
+   "sacks_on_box_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 41.0,
+      "log_step": null,
+      "text_zh": "观察：全程未见袋内有汉堡、也未见袋被放到储物箱顶面；推断：评分器 q_score=0 与此一致。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "1",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hamburgers_in_sacks_1",
+   "where": "操作中",
+   "t": 22.0,
+   "log_step": null,
+   "evidence_zh": "观察：视频约 21-40 s 机器人抓着汉堡搬运，之后没有任何汉堡入袋；评分器 q_score=0。推断：卡在第1个汉堡入袋这一步。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "视频为头/腕相机，汉堡与纸袋最终位置、汉堡放开的时刻不清楚，无状态日志；无法确认具体可见现象；agent 自述袋口位置 IK 不可达仅为会话证据，不据此定根因。",
+  "unmet": {
+   "hamburgers_in_sacks_1": false,
+   "hamburgers_in_sacks_2": false,
+   "hamburgers_in_sacks_3": false,
+   "sacks_on_box_1": false,
+   "sacks_on_box_2": false,
+   "sacks_on_box_3": false
+  }
+ },
+ "summary_zh": "观察：机器人抓取汉堡并搬运（视频约 21-40 s），但视频中没有汉堡入袋，6 个节点均未达成，q_score=0。断点为第1个汉堡入袋（操作中）；现象与根因因视角遮挡且无状态日志而留空。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "标注与视频一致：约21-42 s 夹爪抓汉堡搬运，之后未见汉堡入袋，q_score=0；现象/根因留空并有 gap_zh，可接受。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:54:32Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-packing-meal-for-delivery-i00-privileged--6usjiz3/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-packing-meal-for-delivery-i00-privileged--6usjiz3/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "55f2709c28fc7d843c008d15752fab1c7f6f84f8920ddcbe37fd8b69b81463f0",
+   "video_sha256": "856be27b02e50711c51c5349b574edc5659933674d390654ccd2709879e5d87d",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/polishing_shoes/trajectories/behavior-polishing-shoes-i00-privileged--unvsgvw.json b/phaseR/behavior-1k/polishing_shoes/trajectories/behavior-polishing-shoes-i00-privileged--unvsgvw.json
new file mode 100644
index 0000000000000000000000000000000000000000..deefeed4672903de2fdc882270d5d6f7ac8f2019
--- /dev/null
+++ b/phaseR/behavior-1k/polishing_shoes/trajectories/behavior-polishing-shoes-i00-privileged--unvsgvw.json
@@ -0,0 +1,245 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-polishing-shoes-i00-privileged--unvsgvw",
+ "taskKey": "behavior-1k/polishing_shoes",
+ "dag_sha256": "6b512c37aee3d60fb338b2a855d48e655510d37df6d7fc9eeaccf7d625b63a91",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "deterministic": 0,
+   "missing_trajectory": 1,
+   "grader_error": 0,
+   "infra_error": 0,
+   "agent_sessions": 1,
+   "agent_resets": 0,
+   "agent_sim_steps": 0,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 8.597,
+   "peak_memory_gb": 12.779,
+   "peak_processes": 47,
+   "peak_disk_gb": 0.532,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "none",
+  "url": null,
+  "duration_s": null,
+  "fps": null,
+  "frames": null,
+  "sha256": null
+ },
+ "state": null,
+ "progress": {
+  "method": "none",
+  "denominator": 5,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "shoes_clean_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：run.json reward 中 missing_trajectory=1、agent_sim_steps=0、agent_replays=0，没有交付轨迹，也没有视频和状态日志；该节点无法从任何图像或评分字段判定，故 intervals/final 为空。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹。"
+     }
+    ]
+   },
+   "shoes_clean_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：run.json reward 中 missing_trajectory=1、agent_sim_steps=0、agent_replays=0，没有交付轨迹，也没有视频和状态日志；该节点无法从任何图像或评分字段判定，故 intervals/final 为空。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹。"
+     }
+    ]
+   },
+   "shoes_by_stool_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：run.json reward 中 missing_trajectory=1、agent_sim_steps=0、agent_replays=0，没有交付轨迹，也没有视频和状态日志；该节点无法从任何图像或评分字段判定，故 intervals/final 为空。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹。"
+     }
+    ]
+   },
+   "shoes_by_stool_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：run.json reward 中 missing_trajectory=1、agent_sim_steps=0、agent_replays=0，没有交付轨迹，也没有视频和状态日志；该节点无法从任何图像或评分字段判定，故 intervals/final 为空。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹。"
+     }
+    ]
+   },
+   "brush_on_stand": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：run.json reward 中 missing_trajectory=1、agent_sim_steps=0、agent_replays=0，没有交付轨迹，也没有视频和状态日志；该节点无法从任何图像或评分字段判定，故 intervals/final 为空。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹。"
+     }
+    ]
+   },
+   "brush_in_hand": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：run.json reward 中 missing_trajectory=1、agent_sim_steps=0、agent_replays=0，没有交付轨迹，也没有视频和状态日志；该节点无法从任何图像或评分字段判定，故 intervals/final 为空。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "shoes_clean_1",
+   "t": null,
+   "log_step": null,
+   "where": "首步前",
+   "evidence_zh": "观察：奖励记录 missing_trajectory=1，agent_sim_steps=0；日志（log step 2-73，agent 工作约 7200 s 预算）只是读源码、启动后台模拟器驱动并在后期探测第2只鞋摆放（log step 64、68），最终没有保存或交付任何动作轨迹，因此没有被评分的尝试。"
+  },
+  "unmet": {
+   "shoes_clean_1": null,
+   "shoes_clean_2": null,
+   "shoes_by_stool_1": null,
+   "shoes_by_stool_2": null,
+   "brush_on_stand": null
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无视频、无状态日志、无交付轨迹；agent 因自身原因未交付（预算用尽，最后几步为空输出），按规定不给现象与根因。"
+ },
+ "summary_zh": "观察：本次运行没有交付任何轨迹（missing_trajectory=1），无视频和状态日志，所有节点无法判定（intervals/final 为空）。agent 把约 7200 s 的会话用于读源码和在后台模拟器中调试，直到预算结束都未交付；断点记为首步前，不给现象和根因。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "本次运行无交付轨迹、无视频和状态日志（missing_trajectory=1，agent_sim_steps=0），节点全部 unknown、断点首步前、现象与根因为空并附 gap_zh，符合 03 第5条和 D23 的规则；未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:53:52Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-polishing-shoes-i00-privileged--unvsgvw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-polishing-shoes-i00-privileged--unvsgvw/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "60e910ea810eec3c1954820c346ff0073a16c46c86283341a3fe63338bebd5c2",
+   "video_sha256": null,
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; no local video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/polishing_shoes/trajectories/behavior-polishing-shoes-i00-standard--bhahyym.json b/phaseR/behavior-1k/polishing_shoes/trajectories/behavior-polishing-shoes-i00-standard--bhahyym.json
new file mode 100644
index 0000000000000000000000000000000000000000..ada6c76749fab4ba91c7d80dbe4d180e0eea1de1
--- /dev/null
+++ b/phaseR/behavior-1k/polishing_shoes/trajectories/behavior-polishing-shoes-i00-standard--bhahyym.json
@@ -0,0 +1,329 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-polishing-shoes-i00-standard--bhahyym",
+ "taskKey": "behavior-1k/polishing_shoes",
+ "dag_sha256": "6b512c37aee3d60fb338b2a855d48e655510d37df6d7fc9eeaccf7d625b63a91",
+ "outcome": {
+  "success": 0,
+  "reward": 0.6666666666666666,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 44983,
+   "sim_steps": 44983,
+   "physics_steps": 179932,
+   "steps": 28051,
+   "elapsed_s": 9506.0,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "replay_error": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "q_score": 0.6666666666666666,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-polishing-shoes-i00-standard--bhahyym/replay.mp4",
+  "duration_s": 1499.434,
+  "fps": 30.0,
+  "frames": 44983,
+  "sha256": "2163bea9b7ae1465126bd3bab79697e0ee35d172ff23cbaad34445349db9a659"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 5,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "shoes_clean_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 1499,
+      "log_step": null,
+      "text_zh": "观察：腕部画面分辨不出鞋面上的细小灰粒；刷子是否清除了灰尘无法从画面确认；agent 会话中后期仍在反复刷同一只鞋（log step 348-361），但这只是自述，不作节点依据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "双臂协作",
+      "how": "抓握",
+      "note_zh": "一臂夹鞋、左臂持刷，刷毛擦过鞋壁并旋转鞋（视频约 1000-1450 s，log step 288-361）；放地后持刷擦也出现过（约视频 400-640 s）。"
+     }
+    ]
+   },
+   "shoes_clean_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 1499,
+      "log_step": null,
+      "text_zh": "观察：腕部画面分辨不出鞋面上的细小灰粒；刷子是否清除了灰尘无法从画面确认；agent 会话中后期仍在反复刷同一只鞋（log step 348-361），但这只是自述，不作节点依据。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "双臂协作",
+      "how": "抓握",
+      "note_zh": "一臂夹鞋、左臂持刷，刷毛擦过鞋壁并旋转鞋（视频约 1000-1450 s，log step 288-361）；放地后持刷擦也出现过（约视频 400-640 s）。"
+     }
+    ]
+   },
+   "shoes_by_stool_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      745,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 745,
+      "log_step": 173,
+      "text_zh": "观察：视频约 740 s 起第一只鞋放在布面脚凳边（log step 173 释放）；其后为清洁被拿起再放回，终态鞋仍在凳边（视频 1499 s，log step 364 放回、测得两鞋间距约 0.01 m）。时间为粗估。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右夹爪夹鞋，搬到脚凳边放下。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘载鞋驱动到凳边。"
+     }
+    ]
+   },
+   "shoes_by_stool_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      790,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 790,
+      "log_step": 238,
+      "text_zh": "观察：log step 238 第二只鞋放在第一只旁；视频 905 s 帧可见两只黑鞋在脚凳前；推断：约 790 s 起成立（粗估）。视频 1246 s（step 338）和结尾（step 364）再次放回，同样紧邻。视频结束时第二只鞋尚未被再次夹起（log step 366 刚开始抓）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右夹爪夹鞋，放到第一只鞋旁、脚凳边。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘载鞋驱动到放置点。"
+     }
+    ]
+   },
+   "brush_on_stand": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      133
+     ],
+     [
+      842,
+      868
+     ],
+     [
+      1304,
+      1312
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": null,
+      "text_zh": "观察：视频 0 s 起刷子在台灯旁深色支架上（初态成立）。"
+     },
+     {
+      "t": 135,
+      "log_step": 56,
+      "text_zh": "观察：视频约 133-138 s 左夹爪夹起刷子，节点失去。"
+     },
+     {
+      "t": 842,
+      "log_step": 276,
+      "text_zh": "观察：视频 842 s 头部画面可见红刷放在支架顶面（log step 276 释放，control_steps 25213/30=840 s）；约 870 s 重新抓起（log step 283），时间不确定。"
+     },
+     {
+      "t": 1304,
+      "log_step": 342,
+      "text_zh": "推断：log step 342 第二次放刷（control_steps 39116/30=1304 s），不久又被重新抓起（log step 346）。"
+     },
+     {
+      "t": 1499,
+      "log_step": null,
+      "text_zh": "观察：视频末帧左腕画面仍持刷，刷子不在支架上，final 为 false。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "左夹爪握刷，两次把刷放回支架后又抓起，终态仍持刷。"
+     }
+    ]
+   },
+   "brush_in_hand": {
+    "method": "visual",
+    "intervals": [
+     [
+      135,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 135,
+      "log_step": 56,
+      "text_zh": "观察：视频 126-138 s 左腕画面中红色刷子由远变近、被夹爪夹住并随臂移动（log step 56 左夹爪合拢提刷）；之后左腕画面一直有红刷。视频 836-846 s 刷子一度放在架上，约 870 s 重新抓起（log step 276、283）；事件一旦达成保持已达。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "左二指夹爪夹住红色刷柄把刷子拿离深色支架，之后一直持刷驱动（视频约 135 s 起）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘载着机械臂和持刷驱动到鞋旁。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.6666666666666666,
+   "unit": "比例",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "shoes_clean_2",
+   "where": "操作中",
+   "t": 1499,
+   "evidence_zh": "观察：视频全程机器人持刷反复擦鞋、搬鞋、放回，至 1499 s 仍在对第二只鞋做清洁（log step 366 刚夹起鞋）；评分器 q_score=0.667<1，success=0。推断：灰尘未清完或未被评分器认可；具体哪一鞋仍有灰无法从画面判定。刷子也未放回支架（视频末帧左腕仍持刷）。",
+   "log_step": 366
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无状态日志；画面分辨不出鞋上残留灰粒，故两个清洁节点 unknown，现象和根因无法证明，留空。run 以 token 预算用完结束（log step 367，agent 侧原因，不作现象）。视频时间由 log 的 control_steps/30 与视频帧核对（视频 840 s 刷放支架），其余为粗估。",
+  "unmet": {
+   "shoes_clean_1": null,
+   "shoes_clean_2": null,
+   "shoes_by_stool_1": true,
+   "shoes_by_stool_2": true,
+   "brush_on_stand": false
+  }
+ },
+ "summary_zh": "观察：机器人约视频 135 s 起持刷，约 745-790 s 把两只鞋放到布面脚凳边，随后为清洁反复搬鞋、擦鞋并放回，视频 1499 s 仍在处理，刷子终态仍在手中未放回支架。q_score=0.667，success=0；两个清洁节点因无法看清灰粒而 unknown，刷在架节点 final false，靠凳两节点 true（粗估）。断点取清洁第2只鞋；无状态日志，现象与根因无证据留空，run 因 token 预算耗尽结束。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "抽查画面与 log 后未发现必须修改的错误：刷在架区间（842、1306 s 头部画面可见刷在架顶，1318 s 已持刷）、靠凳节点、两个清洁节点 unknown（无状态日志、腕部画面分不清灰粒、q_score=2/3 的分母未知，无法判定）、断点与空现象层均有证据支撑，gap_zh 已说明原因，token 预算耗尽未被当作现象。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:57:49Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-polishing-shoes-i00-standard--bhahyym/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-polishing-shoes-i00-standard--bhahyym/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6212d57079955927ae8eb40b592d3cf93b54a852d4a16bb61670368c0d332437",
+   "video_sha256": "2163bea9b7ae1465126bd3bab79697e0ee35d172ff23cbaad34445349db9a659",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/preparing_lunch_box/trajectories/behavior-preparing-lunch-box-i00-privileged--j5x4svs.json b/phaseR/behavior-1k/preparing_lunch_box/trajectories/behavior-preparing-lunch-box-i00-privileged--j5x4svs.json
new file mode 100644
index 0000000000000000000000000000000000000000..a6355b2dac32a835573692afc06e6c377fa5868f
--- /dev/null
+++ b/phaseR/behavior-1k/preparing_lunch_box/trajectories/behavior-preparing-lunch-box-i00-privileged--j5x4svs.json
@@ -0,0 +1,336 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-preparing-lunch-box-i00-privileged--j5x4svs",
+ "taskKey": "behavior-1k/preparing_lunch_box",
+ "dag_sha256": "13ccee1b01706e6c1e54d733c8a0aad46f4334041c109c5ec7a4cb7681fcab72",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 1453,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 1603,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 1005.5,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 3,
+   "agent_resets": 1,
+   "agent_sim_steps": 0,
+   "agent_records": 1,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-preparing-lunch-box-i00-privileged--j5x4svs.mp4",
+  "duration_s": 53.434,
+  "fps": 30.0,
+  "frames": 1603,
+  "sha256": "d3beccef9c1c43417ea3f60414af85a8cd4072dcb983e270057c21f353b3ade9"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 7,
+  "base": null,
+  "peak": null,
+  "final": 1,
+  "nodes": {
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 53.0,
+      "log_step": 63,
+      "text_zh": "观察：整段视频未见机器人接近冰箱或门板离开柜体；agent 脚本也未包含开冰箱动作（log step 63 输出只涉及苹果、推曲奇、三明治）。"
+     }
+    ]
+   },
+   "apple_halves_in_box_1": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 53.0,
+      "log_step": 63,
+      "text_zh": "观察：视频 5-53 s 的头部/腕部画面中，苹果、三明治、曲奇始终留在砧板上，未见任何一件进入打包盒；结尾时头相机朝向天花板和窗外。推断：评分 q_score=0，与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "log step 63：脚本令左、右夹爪抓苹果半块，输出 grasp ... False；视频中苹果仍留在砧板上，未见带起。"
+     }
+    ]
+   },
+   "apple_halves_in_box_2": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 53.0,
+      "log_step": 63,
+      "text_zh": "观察：视频 5-53 s 的头部/腕部画面中，苹果、三明治、曲奇始终留在砧板上，未见任何一件进入打包盒；结尾时头相机朝向天花板和窗外。推断：评分 q_score=0，与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "同上，右夹爪抓第二块苹果，输出 False。"
+     }
+    ]
+   },
+   "sandwich_in_box": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 53.0,
+      "log_step": 63,
+      "text_zh": "观察：视频 5-53 s 的头部/腕部画面中，苹果、三明治、曲奇始终留在砧板上，未见任何一件进入打包盒；结尾时头相机朝向天花板和窗外。推断：评分 q_score=0，与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "log step 63：push_edge(SW…) stroke 0-3，三明治约移动 23 cm（x 4.23→4.46）。"
+     },
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "log step 63：边缘夹取 pinch club_sandwich False，未成功入盒。"
+     }
+    ]
+   },
+   "cookie_in_box": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 53.0,
+      "log_step": 63,
+      "text_zh": "观察：视频 5-53 s 的头部/腕部画面中，苹果、三明治、曲奇始终留在砧板上，未见任何一件进入打包盒；结尾时头相机朝向天花板和窗外。推断：评分 q_score=0，与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "log step 63 无曲奇操作输出；视频中曲奇留在砧板上，未见夹爪接触"
+     }
+    ]
+   },
+   "tea_bottle_in_box": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 53.0,
+      "log_step": 63,
+      "text_zh": "观察：视频中从未打开冰箱，茶瓶未被取出。"
+     }
+    ]
+   },
+   "fridge_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": 63,
+      "text_zh": "推断：冰箱初始关闭（DAG 说明），视频中机器人始终未去冰箱，故按条件本身 0 s 起成立至结尾；冰箱本身未出现在画面里，未直接看到。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "无量纲",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "apple_halves_in_box_1",
+   "t": 5.0,
+   "log_step": 63,
+   "where": "操作中",
+   "evidence_zh": "观察：log step 63 的脚本输出，两块苹果 grasp 均为 False，最终 inside 全部 False；视频 5-53 s 苹果和其他物品始终在砧板上。推断：最先尝试的苹果入盒从未完成，其后的推/夹三明治也未成功。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 5.0,
+    "log_step": 63,
+    "evidence_zh": "观察：log step 63 输出两次 grasp ... False，视频中苹果仍在砧板上，夹爪未带起任何物体。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "grasp 失败原因不能由现有证据判定：step 46 同样的 move_base 停滞（0.17 m short）下抓取成功，step 63 与其差别（手、下降高度 2 mm、速度）无法区分；无状态日志，视频看不清指尖；冰箱从未入画，fridge_closed 为推断。",
+  "unmet": {
+   "apple_halves_in_box_1": false,
+   "apple_halves_in_box_2": false,
+   "sandwich_in_box": false,
+   "cookie_in_box": false,
+   "tea_bottle_in_box": false,
+   "fridge_closed": true
+  }
+ },
+ "summary_zh": "图示任务：5 件物品放入打包盒并关冰箱。回放视频中没有任何物品入盒（q_score=0），苹果抓取在 log step 63 均失败，其后机器人推/夹三明治也未成功；从未打开冰箱。断在第1块苹果入盒（操作中）；失败根因无法由现有证据判定（step 46 同样的底盘停滞下抓取成功）。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 3,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点值、时间线、断点（第1块苹果入盒，操作中）和 final=1（fridge_closed 起始即真，q=0 一致）均与视频、log 一致；唯一问题是根因：「抓取和接触设计」只靠 agent 自述，且与 log 中同一 agent 的其他读数冲突，证据不足。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点值、断点（第1块苹果入盒，操作中）与 final=1 与视频/log 一致；但根因「轨迹或逆解（到位偏差 0.16 m）」被同会话 step 46 反驳（同样的 0.17 m 停滞下苹果抓取成功），且 cookie_in_box 的 contact_actual 把推/夹三明治的读数记到了曲奇上。",
+    "n_issues": 2,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 3,
+    "kind": "confirm",
+    "verdict": "agree",
+    "summary_zh": "两个 round-2 问题的修改均已正确落实：root_causes 置空且 gap_zh/summary/conf_reason 同步；曲奇接触改为未知，三明治改为推/扫+抓握，断点与 summary 已无「推曲奇」。",
+    "n_issues": 0,
+    "n_checks": 2
+   }
+  ],
+  "open_zh": [],
+  "confirm": {
+   "verdict": "confirmed",
+   "summary_zh": "两个 round-2 问题的修改均已正确落实：root_causes 置空且 gap_zh/summary/conf_reason 同步；曲奇接触改为未知，三明治改为推/扫+抓握，断点与 summary 已无「推曲奇」。",
+   "checks_ok": "2/2",
+   "n_factual_errors": 0
+  }
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:54Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-preparing-lunch-box-i00-privileged--j5x4svs/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-preparing-lunch-box-i00-privileged--j5x4svs/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-preparing-lunch-box-i00-privileged--j5x4svs/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-preparing-lunch-box-i00-privileged--j5x4svs/04_review_r2"
+   },
+   {
+    "stage": "reconcile",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-preparing-lunch-box-i00-privileged--j5x4svs/05_reconcile_r2"
+   },
+   {
+    "stage": "confirm",
+    "round": 3,
+    "ok": true,
+    "space": "work/runs/behavior-preparing-lunch-box-i00-privileged--j5x4svs/06_confirm_r3"
+   }
+  ],
+  "confirm_prompt": "prompts/06_confirm.md",
+  "confirm_prompt_sha256": "b9f56c095f7ea5e609e58bb21e076f565766c370333d71c7a3512cffb4d78267",
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ef56e536d73a864a58f4f541430c3ef96513d252152474f3ddbfb553134f654f",
+   "video_sha256": "d3beccef9c1c43417ea3f60414af85a8cd4072dcb983e270057c21f353b3ade9",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ },
+ "conf_reason_zh": "根因无法判定，root_causes 为空；断点与现象有 log 与视频支持。"
+}
diff --git a/phaseR/behavior-1k/put_together_a_basic_pruning_kit/trajectories/behavior-put-together-a-basic-pruning-kit-i00-standard--v2x6k9g.json b/phaseR/behavior-1k/put_together_a_basic_pruning_kit/trajectories/behavior-put-together-a-basic-pruning-kit-i00-standard--v2x6k9g.json
new file mode 100644
index 0000000000000000000000000000000000000000..b0a3b17434f38d6ddfcdf85b164ae5e9df1ee6de
--- /dev/null
+++ b/phaseR/behavior-1k/put_together_a_basic_pruning_kit/trajectories/behavior-put-together-a-basic-pruning-kit-i00-standard--v2x6k9g.json
@@ -0,0 +1,243 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-put-together-a-basic-pruning-kit-i00-standard--v2x6k9g",
+ "taskKey": "behavior-1k/put_together_a_basic_pruning_kit",
+ "dag_sha256": "45588bc2f9af512dc48078f6ad004600b8d7191ff8df999c957949771ce64a79",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 2625,
+   "sim_steps": 2775,
+   "physics_steps": 11100,
+   "steps": 410,
+   "elapsed_s": 6474.8,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 150,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 0.0,
+   "goal_first_step": 0,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.286,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.357,
+   "peak_cpus": 1.28,
+   "peak_pids": 245,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-put-together-a-basic-pruning-kit-i00-standard--v2x6k9g.mp4",
+  "duration_s": 92.6,
+  "fps": 10.0,
+  "frames": 926,
+  "sha256": "5765c173688c138f01a35aea94697e3e431737195715a06beab2c50b407d5f31"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 5,
+  "base": 2,
+  "peak": 2,
+  "final": 2,
+  "nodes": {
+   "toolbox_on_floor": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 92.4,
+      "log_step": null,
+      "text_zh": "观察：视频 0 s 起红色工具箱在车库地面，视频 92.4 s 末帧仍在地面（头部相机）；中间腕部相机常被红色箱体占满，未见箱体被带离地面的清晰证据。"
+     }
+    ]
+   },
+   "toolbox_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 92.4,
+      "log_step": null,
+      "text_zh": "观察：视频 8–92 s 头部相机所见箱体始终为闭盖红色扁盒，未见盖掀开露出内腔；末帧 92.4 s 仍闭合。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "观察：视频约 32–90 s 腕部相机反复被红色箱体占满、夹爪贴近箱体，但盖未被翻开；具体接触方式无法从画面确认。"
+     }
+    ]
+   },
+   "pruner_in_toolbox": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 92.4,
+      "log_step": null,
+      "text_zh": "观察：视频末帧箱内无剪具，且箱盖未开；q_score=0。"
+     }
+    ]
+   },
+   "shears_in_toolbox": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 92.4,
+      "log_step": null,
+      "text_zh": "观察：同上，箱盖始终闭合，大剪刀未入箱。"
+     }
+    ]
+   },
+   "toolbox_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 92.4,
+      "log_step": null,
+      "text_zh": "观察：箱盖从开局到视频 92.4 s 一直闭合（盖从未掀开）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "比例",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "toolbox_opened",
+   "t": 32,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 32–90 s 机器人反复贴近红色工具箱（腕部相机被红色占满、箱体在头部相机中被移近），但盖始终闭合，开箱步骤从未修复；92.4 s 末帧仍闭盖。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "视频只显示三路相机轮换的低分辨率画面，无状态日志，无法判定夹爪与盖的接触细节，故现象与根因留空；agent 日志（log step 108、216、276）显示其多次尝试用不同脚本开盖，属会话证据，未作为现象依据。",
+  "unmet": {
+   "toolbox_on_floor": true,
+   "pruner_in_toolbox": false,
+   "shears_in_toolbox": false,
+   "toolbox_closed": true
+  }
+ },
+ "summary_zh": "观察：视频 92.6 s 内工具箱始终闭盖并在地面，打开工具箱从未达成，两件剪具均未入箱（q_score=0）。推断：断在打开工具箱一步（操作中）。主要证据缺口：无状态日志、画面低清，接触细节与失败原因无法确认。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 5,
+    "summary_zh": "视频只显示工具箱始终闭盖、剪具未入箱，q_score=0 与标注一致；无状态日志，现象与根因留空并有 gap_zh，可接受。未发现可证实的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:54:50Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-put-together-a-basic-pruning-kit-i00-standard--v2x6k9g/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-put-together-a-basic-pruning-kit-i00-standard--v2x6k9g/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6f2efbf8988616a24514a1dd22a0d91c87f524cb29dcbe37dfa39b516d68a4a7",
+   "video_sha256": "5765c173688c138f01a35aea94697e3e431737195715a06beab2c50b407d5f31",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/putting_dirty_dishes_in_sink/trajectories/behavior-putting-dirty-dishes-in-sink-i00-privileged--dknhpre.json b/phaseR/behavior-1k/putting_dirty_dishes_in_sink/trajectories/behavior-putting-dirty-dishes-in-sink-i00-privileged--dknhpre.json
new file mode 100644
index 0000000000000000000000000000000000000000..038cab200516094b298b13f5bf9b14a4ff33d767
--- /dev/null
+++ b/phaseR/behavior-1k/putting_dirty_dishes_in_sink/trajectories/behavior-putting-dirty-dishes-in-sink-i00-privileged--dknhpre.json
@@ -0,0 +1,214 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-putting-dirty-dishes-in-sink-i00-privileged--dknhpre",
+ "taskKey": "behavior-1k/putting_dirty_dishes_in_sink",
+ "dag_sha256": "0ae83395c13ee3df9019df559327fe10e26a960f6acd108d99bcbe4e135e8d04",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "deterministic": 0,
+   "missing_trajectory": 1,
+   "grader_error": 0,
+   "infra_error": 0,
+   "agent_sessions": 1,
+   "agent_resets": 2,
+   "agent_sim_steps": 0,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.351,
+   "peak_memory_gb": 21.915,
+   "peak_processes": 78,
+   "peak_disk_gb": 0.619,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "none",
+  "url": null,
+  "duration_s": null,
+  "fps": null,
+  "frames": null,
+  "sha256": null
+ },
+ "state": null,
+ "progress": {
+  "method": "none",
+  "denominator": 4,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "bowls_in_sink_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 117,
+      "text_zh": "观察：无视频（video.kind none），无状态日志；run.json 的 reward 为 missing_trajectory=1、agent_records=0、agent_replays=0，没有交付轨迹，故无法判断该节点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹（missing_trajectory=1），agent 仅在开发会话中试验，未进入被评分的 episode。"
+     }
+    ]
+   },
+   "bowls_in_sink_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 117,
+      "text_zh": "观察：无视频（video.kind none），无状态日志；run.json 的 reward 为 missing_trajectory=1、agent_records=0、agent_replays=0，没有交付轨迹，故无法判断该节点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹（missing_trajectory=1），agent 仅在开发会话中试验，未进入被评分的 episode。"
+     }
+    ]
+   },
+   "plates_in_sink_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 117,
+      "text_zh": "观察：无视频（video.kind none），无状态日志；run.json 的 reward 为 missing_trajectory=1、agent_records=0、agent_replays=0，没有交付轨迹，故无法判断该节点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹（missing_trajectory=1），agent 仅在开发会话中试验，未进入被评分的 episode。"
+     }
+    ]
+   },
+   "plates_in_sink_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 117,
+      "text_zh": "观察：无视频（video.kind none），无状态日志；run.json 的 reward 为 missing_trajectory=1、agent_records=0、agent_replays=0，没有交付轨迹，故无法判断该节点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹（missing_trajectory=1），agent 仅在开发会话中试验，未进入被评分的 episode。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": null,
+   "unit": "比例",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bowls_in_sink_1",
+   "t": null,
+   "log_step": 117,
+   "where": "首步前",
+   "evidence_zh": "观察：reward 中 missing_trajectory=1、agent_records=0、agent_replays=0；log step 114-117 agent 在开发会话中机器人两次在窄通道摔倒，最后一步（log step 117）只写了 cmd15.py，未交付任何可回放的轨迹。被评分的 episode 里没有任何餐具操作。"
+  },
+  "unmet": {
+   "bowls_in_sink_1": null,
+   "bowls_in_sink_2": null,
+   "plates_in_sink_1": null,
+   "plates_in_sink_2": null
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无视频、无状态日志；未交付轨迹（agent 侧原因，会话在约 7200 s 的预算内结束，仍未录制 run-oracle 轨迹）。会话试验（log step 114-117 的通道摔倒、step 99 的边缘夹取）只属于会话证据，不构成现象；故现象与根因均留空。"
+ },
+ "summary_zh": "观察：agent 在会话中调试抓取、过门和通道导航（log step 99-117），始终没有录制并交付轨迹（missing_trajectory=1），被评分的 episode 无内容，四个节点因无视频无状态日志均为 unknown。断点记在第1只碗入槽的首步前；现象和根因因缺证据留空。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "无视频无状态日志、未交付轨迹（missing_trajectory=1），四节点 unknown、现象与根因留空并写 gap_zh，符合 OPEN-16/D23；断点记为首步前合理，未把会话中的摔倒当作被评分 episode 的现象。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:54:23Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-putting-dirty-dishes-in-sink-i00-privileged--dknhpre/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-putting-dirty-dishes-in-sink-i00-privileged--dknhpre/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "d54d6b6c4fd4d9d93a650ccc89321eeba759dbf25cbdf26742ccd16115be1577",
+   "video_sha256": null,
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; no local video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/re_shelving_library_books/trajectories/behavior-re-shelving-library-books-i00-privileged--blm2bba.json b/phaseR/behavior-1k/re_shelving_library_books/trajectories/behavior-re-shelving-library-books-i00-privileged--blm2bba.json
new file mode 100644
index 0000000000000000000000000000000000000000..61cdd07c01f2cd83bfcf953d464387691bc65152
--- /dev/null
+++ b/phaseR/behavior-1k/re_shelving_library_books/trajectories/behavior-re-shelving-library-books-i00-privileged--blm2bba.json
@@ -0,0 +1,249 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-re-shelving-library-books-i00-privileged--blm2bba",
+ "taskKey": "behavior-1k/re_shelving_library_books",
+ "dag_sha256": "180905a042e2c1a91f574fec6849e3d47fa606674a84c78c7b0e7b202d92a7fa",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 1134,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 0,
+   "goal_first_step": -1,
+   "sim_steps": 1134,
+   "fallen": 1,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 591.0,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 20,
+   "agent_resets": 0,
+   "agent_sim_steps": 7072,
+   "agent_records": 12,
+   "agent_replays": 1,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.247,
+   "peak_memory_gb": 15.451,
+   "peak_processes": 78,
+   "peak_disk_gb": 5.141,
+   "peak_cpus": 8.09,
+   "peak_pids": 584,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-re-shelving-library-books-i00-privileged--blm2bba.mp4",
+  "duration_s": 37.8,
+  "fps": 30.0,
+  "frames": 1134,
+  "sha256": "c1f20211c7df6d91a98e739eb70bc8b8f7d3c15cc9d8a8dafa5fb81aba780b85"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "books_in_bookcases_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 37.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-37.8 s 全程桌上的蓝色和绿色书一直留在桌面，没有任何一本被夹起或进入白色书柜格口；评分器 goal=0、q_score=0.0 与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：机械臂的夹爪在桌面书附近压/推书，书未被抓握抬起；视角混乱，具体接触只能大致判断。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在桌与椅之间移动。"
+     }
+    ]
+   },
+   "books_in_bookcases_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 37.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-37.8 s 全程桌上的蓝色和绿色书一直留在桌面，没有任何一本被夹起或进入白色书柜格口；评分器 goal=0、q_score=0.0 与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：机械臂的夹爪在桌面书附近压/推书，书未被抓握抬起；视角混乱，具体接触只能大致判断。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在桌与椅之间移动。"
+     }
+    ]
+   },
+   "books_in_bookcases_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 37.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-37.8 s 全程桌上的蓝色和绿色书一直留在桌面，没有任何一本被夹起或进入白色书柜格口；评分器 goal=0、q_score=0.0 与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：机械臂的夹爪在桌面书附近压/推书，书未被抓握抬起；视角混乱，具体接触只能大致判断。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在桌与椅之间移动。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "books_in_bookcases_1",
+   "t": 5.0,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频中书始终在桌上、未被抬起；视频后段相机视角倾斜翻转，评分器 fallen=1。agent 最终报告（log 最后一步）自述桌被椅子和楼梯围住，书 2/3 够不到，推书的压/拖没成功，第一本书的拾取尚未完成。"
+  },
+  "unmet": {
+   "books_in_bookcases_1": false,
+   "books_in_bookcases_2": false,
+   "books_in_bookcases_3": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 5.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频全程没有任何一本书被夹爪夹住抬起，书留在桌上；agent 自述（最后一步）压住书封面拖动时书只移动≤5 cm。"
+   },
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 20.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频后段视角倾斜、天花板/墙面占满画面，评分器字段 fallen=1，机器人翻倒；具体翻倒时刻在视频中难以精确定位。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "抓取和接触设计",
+    "alt_feasible": "未知",
+    "evidence_zh": "agent 自述：桌被椅子和楼梯围住，只有南北两端可进，书2、3超出臂展，推/压书的方式没抓住书，书移动≤5 cm；未验证是否有可行的取书方案。",
+    "source_zh": "agent 自述"
+   }
+  ],
+  "gap_zh": "无状态日志，视频相机视角混乱，翻倒的精确时刻和夹爪与书的具体接触无法确定；根因来自 agent 自述，待核实。"
+ },
+ "summary_zh": "观察：三本书全程留在桌上，没有一本进入书柜，3 个节点均未达成（评分器 goal=0，q_score=0）。断点在第1本书入书柜的取书/放入步骤：夹爪没能抓起书，视频后段机器人翻倒（fallen=1）。根因为 agent 自述的取书接触方案不可行，证据主要来自自述，视频视角混乱。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 5,
+    "summary_zh": "三个书柜节点均未达成与评分器 goal=0、q_score=0、视频中书始终留在桌上一致；断点在第1本书（agent 称只有它可达）、没建立接触 + 平衡（fallen=1）以及 agent 自述的抓取和接触设计根因都有支撑且已如实标注 alt_feasible=未知 与低 conf。视频为四宫格、视角混乱，未发现可证伪的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:55:03Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-re-shelving-library-books-i00-privileged--blm2bba/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-re-shelving-library-books-i00-privileged--blm2bba/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "e8ee77fafb4b59de1283600dffa5d777a65ca9a49caa9cd573ab1b347020a1a9",
+   "video_sha256": "c1f20211c7df6d91a98e739eb70bc8b8f7d3c15cc9d8a8dafa5fb81aba780b85",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/rearrange_your_room/trajectories/behavior-rearrange-your-room-i00-standard--btvb8px.json b/phaseR/behavior-1k/rearrange_your_room/trajectories/behavior-rearrange-your-room-i00-standard--btvb8px.json
new file mode 100644
index 0000000000000000000000000000000000000000..c0cbc4023a5f46e31e36323212afecead320d6a8
--- /dev/null
+++ b/phaseR/behavior-1k/rearrange_your_room/trajectories/behavior-rearrange-your-room-i00-standard--btvb8px.json
@@ -0,0 +1,228 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-rearrange-your-room-i00-standard--btvb8px",
+ "taskKey": "behavior-1k/rearrange_your_room",
+ "dag_sha256": "70f0904f1b16483ad518f0a5db5d80d7e537e1442fcfe1c08c6afb0a8657f110",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "graded_by_live": 1,
+   "replays": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 4935,
+   "sim_steps": 4935,
+   "physics_steps": 19740,
+   "steps": 1645,
+   "elapsed_s": 4522.3,
+   "video_rendered": 1,
+   "video_live": 1,
+   "video_pending": 0,
+   "settled_success": 0,
+   "settle_steps": 0,
+   "sim_breakdown": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "q_score": 1.0,
+   "goal_first_step": 4776,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.417,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.301,
+   "peak_cpus": 4.58,
+   "peak_pids": 150,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-rearrange-your-room-i00-standard--btvb8px.mp4",
+  "duration_s": 164.6,
+  "fps": 10.0,
+  "frames": 1646,
+  "sha256": "fee90ae92e08550acdb96cfb3e943b43bf763f0aa325e901fd5a91071e9c3b91"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 3,
+  "final": 3,
+  "nodes": {
+   "tissue_on_stand": {
+    "method": "visual",
+    "intervals": [
+     [
+      159,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 159,
+      "log_step": null,
+      "text_zh": "观察：视频 158 s 左夹爪位于台面上方，之后张开松手；评分器 goal_first_step=4776/4935 步 → 约 159 s（推断，30 Hz，终点 164.6 s）。视频中纸巾盒细节被遮挡。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "左夹爪夹住纸巾盒，从扶手椅取出并携带（期间也抓枕头），放到床边矮柜"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "携盒驱动"
+     }
+    ]
+   },
+   "pillows_on_beds_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      82,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 82,
+      "log_step": null,
+      "text_zh": "观察：视频 80-82 s 头部相机可见右夹爪放下的浅色枕头落在床面上，86 s 后枕头仍在床上；评分器 q_score=1.0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右夹爪夹住枕头前沿，底盘驱动搬运到床边后松手"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "携枕驱动到床旁"
+     }
+    ]
+   },
+   "pillows_on_beds_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      152,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 152,
+      "log_step": null,
+      "text_zh": "观察：视频 144 s 右腕相机见枕头被夹持，150 s 仍在手中，154 s 头部相机见枕头平放在床面（约 152 s 松手）。推断：时间为估计。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右夹爪夹住第二个枕头并搬到床1"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "携枕驱动"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": null,
+ "summary_zh": "成功：三个节点均达成（枕头1约视频 82 s，枕头2约 152 s，纸巾盒约 159 s），评分器 q_score=1.0、success=1。无状态日志，时间来自视频目测和 goal_first_step 推断，纸巾盒放置的细节被相机遮挡。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "成功运行，三个节点达成，时间与 goal_first_step=4776/4935（30 Hz≈159 s）及视频帧相符；接触标签合理，未发现需修改之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:54:46Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-rearrange-your-room-i00-standard--btvb8px/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-rearrange-your-room-i00-standard--btvb8px/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3d757844e60956291b03ccb59718c50802fe29d4c01fa7db69b053bc6a8b430c",
+   "video_sha256": "fee90ae92e08550acdb96cfb3e943b43bf763f0aa325e901fd5a91071e9c3b91",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/rearranging_kitchen_furniture/trajectories/behavior-rearranging-kitchen-furniture-i00-privileged--n7bl8zk.json b/phaseR/behavior-1k/rearranging_kitchen_furniture/trajectories/behavior-rearranging-kitchen-furniture-i00-privileged--n7bl8zk.json
new file mode 100644
index 0000000000000000000000000000000000000000..62adbd7529a9902ac46e59cc98f1b98d69fde5e4
--- /dev/null
+++ b/phaseR/behavior-1k/rearranging_kitchen_furniture/trajectories/behavior-rearranging-kitchen-furniture-i00-privileged--n7bl8zk.json
@@ -0,0 +1,221 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-rearranging-kitchen-furniture-i00-privileged--n7bl8zk",
+ "taskKey": "behavior-1k/rearranging_kitchen_furniture",
+ "dag_sha256": "c2cd92aef5b1a32fe7dfa519e8e28b4115648494e5b651c881fe0bce10664da7",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 0,
+   "n_actions": 60,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 210,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 701.8,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 3,
+   "agent_sim_steps": 0,
+   "agent_records": 1,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.56,
+   "peak_memory_gb": 13.25,
+   "peak_processes": 46,
+   "peak_disk_gb": 0.892,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-rearranging-kitchen-furniture-i00-privileged--n7bl8zk.mp4",
+  "duration_s": 7.0,
+  "fps": 30.0,
+  "frames": 210,
+  "sha256": "df3212281d7ac633560d9d445e179ba8985780c722c30cb7ae94cd135262ce70"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 5,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cabinet_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 3.0,
+      "log_step": 54,
+      "text_zh": "观察：视频 0–7 s 逐秒采样，机器人静止不动，三件物品和柜门均无变化（评分轨迹是 60 步保持动作，log step 54-55 / run.json n_actions=60）；q_score=0。"
+     }
+    ]
+   },
+   "toaster_in_cabinet": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 3.0,
+      "log_step": 54,
+      "text_zh": "观察：视频 0–7 s 逐秒采样，机器人静止不动，三件物品和柜门均无变化（评分轨迹是 60 步保持动作，log step 54-55 / run.json n_actions=60）；q_score=0。"
+     }
+    ]
+   },
+   "processor_in_cabinet": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 3.0,
+      "log_step": 54,
+      "text_zh": "观察：视频 0–7 s 逐秒采样，机器人静止不动，三件物品和柜门均无变化（评分轨迹是 60 步保持动作，log step 54-55 / run.json n_actions=60）；q_score=0。"
+     }
+    ]
+   },
+   "press_in_cabinet": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 3.0,
+      "log_step": 54,
+      "text_zh": "观察：视频 0–7 s 逐秒采样，机器人静止不动，三件物品和柜门均无变化（评分轨迹是 60 步保持动作，log step 54-55 / run.json n_actions=60）；q_score=0。"
+     }
+    ]
+   },
+   "cabinet_closed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 3.0,
+      "log_step": 54,
+      "text_zh": "观察：视频 0–7 s 逐秒采样，机器人静止不动，三件物品和柜门均无变化（评分轨迹是 60 步保持动作，log step 54-55 / run.json n_actions=60）；q_score=0。 推断：该节点需前三物到位后才成立，故未达成。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "ratio",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "cabinet_opened",
+   "t": 0.0,
+   "log_step": 54,
+   "where": "首步前",
+   "evidence_zh": "观察：视频 0–7 s 机器人原地不动、柜门始终关闭；log step 54-55 显示 agent 在剩约 55 s 时只保存了 60 步保持动作的最小轨迹，完整生成脚本 gen.py 从未录制运行（log step 55）。"
+  },
+  "unmet": {
+   "toaster_in_cabinet": false,
+   "processor_in_cabinet": false,
+   "press_in_cabinet": false,
+   "cabinet_closed": false
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 54,
+    "evidence_zh": "观察：交付的评分轨迹只含保持动作，视频中机器人未移动、未接触柜门或物品，第一步（开柜门）从未开始。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "根因层为空：不能从没尝试或缺交付反推根因；agent 把会话时间用于探索与门把手抓取测试（log step 18-49），未录制可回放的轨迹，但这属于用尽预算，不是根因证据。无状态日志，节点由视频判断。"
+ },
+ "summary_zh": "观察：评分轨迹仅为 60 步保持动作，视频 0–7 s 机器人静止，三件物品无一入柜，5 个节点均未达成（0/5，q_score=0）。断点在 打开柜门 之前（首步前），现象为没尝试；agent 把时间花在探索和试验上，最后约 55 s 才存最小轨迹（log step 54），根因证据不足而留空。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "视频 0–7 s 机器人原地静止，柜门与三件物品无变化，5 个节点均未达成，与 q_score=0、n_actions=60 一致；首步前 / 没尝试、两层根因为空并附 gap_zh，符合 03 的规则（FAIL-4/22）。未发现需要修改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:54:48Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-rearranging-kitchen-furniture-i00-privileged--n7bl8zk/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-rearranging-kitchen-furniture-i00-privileged--n7bl8zk/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "dcde7f551b7ab2c6917a3b42c83c227ec6a8a89b1f0caa7eaf249f2340b95d26",
+   "video_sha256": "df3212281d7ac633560d9d445e179ba8985780c722c30cb7ae94cd135262ce70",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/rearranging_kitchen_furniture/trajectories/behavior-rearranging-kitchen-furniture-i00-standard--gj23gra.json b/phaseR/behavior-1k/rearranging_kitchen_furniture/trajectories/behavior-rearranging-kitchen-furniture-i00-standard--gj23gra.json
new file mode 100644
index 0000000000000000000000000000000000000000..a65f56d5662640b959661892b5113ccaa9b356d1
--- /dev/null
+++ b/phaseR/behavior-1k/rearranging_kitchen_furniture/trajectories/behavior-rearranging-kitchen-furniture-i00-standard--gj23gra.json
@@ -0,0 +1,256 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-rearranging-kitchen-furniture-i00-standard--gj23gra",
+ "taskKey": "behavior-1k/rearranging_kitchen_furniture",
+ "dag_sha256": "c2cd92aef5b1a32fe7dfa519e8e28b4115648494e5b651c881fe0bce10664da7",
+ "outcome": {
+  "success": 0,
+  "reward": 0.25,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "replay_success": 0,
+   "deterministic": 1,
+   "live_matches_replay": 1,
+   "finalized": 1,
+   "missing_log": 0,
+   "n_actions": 25187,
+   "sim_steps": 25187,
+   "physics_steps": 100748,
+   "steps": 8934,
+   "elapsed_s": 9119.7,
+   "video_rendered": 1,
+   "gpu_lost": 0,
+   "replay_error": 0,
+   "instance_ok": 1,
+   "task_failure": 0,
+   "q_score": 0.25,
+   "sim_startups": 1,
+   "resumes": 0,
+   "resume_state_mismatch": 0,
+   "resume_max_tcp_dev_m": 0.0,
+   "resume_max_joint_dev": 0.0,
+   "gave_up": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/behavior-rearranging-kitchen-furniture-i00-standard--gj23gra/replay.mp4",
+  "duration_s": 839.567,
+  "fps": 30.0,
+  "frames": 25187,
+  "sha256": "274e3f4329f016c9b0ceb019af3b8d6472819104ac3e5b75b909104fa20555a7"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 5,
+  "base": null,
+  "peak": null,
+  "final": 2,
+  "nodes": {
+   "cabinet_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      90.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 90.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 90 s 起头部相机可见吊柜门板打开、柜内露出；结束时（视频 780-838 s）柜门仍开着。时间为粗略估计。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "推/扫",
+      "note_zh": "观察：视频中夹爪/手臂推拉柜门；log step 100 自述从门 A 开口伸入推门 B。底盘移动到柜前（底盘×支撑）。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驱动到柜前。"
+     }
+    ]
+   },
+   "toaster_in_cabinet": {
+    "method": "visual",
+    "intervals": [
+     [
+      738.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 738.0,
+      "log_step": 317,
+      "text_zh": "观察：视频约 650-730 s 机器人夹爪抱着烤面包机在柜前；约 740 s 腕部相机可见烤面包机已放在柜内、夹爪退出。agent 自述 log step 317“The toaster is sitting in the cabinet”。推断：q_score=0.25（4 个目标文字中 1 个）与之相符。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频约 650-735 s 夹爪抓住烤面包机抬起、送入柜内并松开。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘移动到柜前（含携物行驶）。"
+     }
+    ]
+   },
+   "processor_in_cabinet": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 170,
+      "text_zh": "观察：视频 780-838 s 画面中没有食品处理机进柜；q_score 0.25 只容许一个物体入柜（推断，且已被烤面包机占用）。log step 170 称食品处理机位置已变。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频中无法确认是否触碰食品处理机。"
+     }
+    ]
+   },
+   "press_in_cabinet": {
+    "method": "visual",
+    "intervals": null,
+    "final": false,
+    "evidence": [
+     {
+      "t": 838.0,
+      "log_step": 361,
+      "text_zh": "观察：视频 745-838 s 夹爪拿着法压壶在柜下方/水槽上方，未进柜；log step 361 末次计划“raise the press in front of the cabinet”，随后 token 预算用尽（log step 362）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频约 760-838 s 夹爪抓住法压壶的把手/壶身抬起。"
+     }
+    ]
+   },
+   "cabinet_closed": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      90.0
+     ]
+    ],
+    "final": false,
+    "evidence": [
+     {
+      "t": 838.0,
+      "log_step": null,
+      "text_zh": "观察：柜门初始关闭（t=0），约 90 s 被打开后直到视频结束（838 s）仍开着，所以终态未关闭。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "press_in_cabinet",
+   "t": 760.0,
+   "log_step": 361,
+   "where": "操作中",
+   "evidence_zh": "观察：烤面包机入柜后，视频约 745-838 s 机器人拿着法压壶在柜前准备插入，视频结束时未入柜、柜门仍开；log step 361 为最后一次执行计划，随后预算用尽（log step 362）。食品处理机也未入柜，但视频上没有对它的操作。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "现象与根因留空：无状态日志，头部相机多处遮挡，视频未显示压壶/处理机具体怎么失败；预算耗尽不是现象。agent 全程大量时间用于逐步规划，但没有证据表明规划本身错误。",
+  "unmet": {
+   "toaster_in_cabinet": true,
+   "processor_in_cabinet": false,
+   "press_in_cabinet": false,
+   "cabinet_closed": false
+  }
+ },
+ "summary_zh": "观察：图形化部分只达到 打开橱柜 与 烤面包机入柜（视频约 738 s），食品处理机与法压壶未入柜，柜门到视频结束仍开着，q_score=0.25。断点在法压壶入柜步骤（视频约 760-838 s 拿着壶在柜前），随后 token 预算用尽。无状态日志，节点时间为视觉粗估；现象与根因因证据不足留空。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 5,
+    "summary_zh": "节点（开柜、烤面包机入柜为真，处理机、压壶未入柜，柜终态未关）与 q_score=0.25、视频所见一致；断点在压壶步骤合理；现象与根因留空且有 gap_zh，因视频与日志无法判定失败机制，可接受。未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:19Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-rearranging-kitchen-furniture-i00-standard--gj23gra/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-rearranging-kitchen-furniture-i00-standard--gj23gra/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f998ee323210df161ae7bed74fc0631ce4e69a3251babf2195c2833341945542",
+   "video_sha256": "274e3f4329f016c9b0ceb019af3b8d6472819104ac3e5b75b909104fa20555a7",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/setting_the_table/trajectories/behavior-setting-the-table-i00-privileged--pipwpla.json b/phaseR/behavior-1k/setting_the_table/trajectories/behavior-setting-the-table-i00-privileged--pipwpla.json
new file mode 100644
index 0000000000000000000000000000000000000000..f512b4d58a06f465f7a3f471987fcdee015443ba
--- /dev/null
+++ b/phaseR/behavior-1k/setting_the_table/trajectories/behavior-setting-the-table-i00-privileged--pipwpla.json
@@ -0,0 +1,334 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-setting-the-table-i00-privileged--pipwpla",
+ "taskKey": "behavior-1k/setting_the_table",
+ "dag_sha256": "6a5f1eba95ccc5ad25db2527a55c61eff05dce14bb67bf276b11bab2a1d89016",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 0,
+   "n_actions": 1583,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 0,
+   "goal_first_step": -1,
+   "sim_steps": 305,
+   "fallen": 1,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 995.4,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 2,
+   "agent_resets": 4,
+   "agent_sim_steps": 0,
+   "agent_records": 8,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 10.017,
+   "peak_memory_gb": 13.33,
+   "peak_processes": 47,
+   "peak_disk_gb": 1.533,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-setting-the-table-i00-privileged--pipwpla.mp4",
+  "duration_s": 10.167,
+  "fps": 30.0,
+  "frames": 305,
+  "sha256": "674118fa1e50038ea540c77cfaeaf5c64ad68604780c78c8ac2d06e82152386c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 10,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cabinet_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-10.2 s 头部相机画面持续翻滚倾斜，只见厨房台面、天花和机械臂，未见橱柜/冰箱门板打开，也未见任何盘子被取出；推断：未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知"
+     }
+    ]
+   },
+   "fridge_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-10.2 s 头部相机画面持续翻滚倾斜，只见厨房台面、天花和机械臂，未见橱柜/冰箱门板打开，也未见任何盘子被取出；推断：未达成。"
+     }
+    ]
+   },
+   "plates_on_table_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程未见盘/蛋糕/餐具被拿起或放上早餐桌（机器人始终在厨房区域）；评分器 q_score=0、goal=0，与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知"
+     }
+    ]
+   },
+   "plates_on_table_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程未见盘/蛋糕/餐具被拿起或放上早餐桌（机器人始终在厨房区域）；评分器 q_score=0、goal=0，与此一致。"
+     }
+    ]
+   },
+   "cupcakes_on_plates_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程未见盘/蛋糕/餐具被拿起或放上早餐桌（机器人始终在厨房区域）；评分器 q_score=0、goal=0，与此一致。"
+     }
+    ]
+   },
+   "cupcakes_on_plates_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程未见盘/蛋糕/餐具被拿起或放上早餐桌（机器人始终在厨房区域）；评分器 q_score=0、goal=0，与此一致。"
+     }
+    ]
+   },
+   "forks_next_to_plates_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程未见盘/蛋糕/餐具被拿起或放上早餐桌（机器人始终在厨房区域）；评分器 q_score=0、goal=0，与此一致。"
+     }
+    ]
+   },
+   "forks_next_to_plates_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程未见盘/蛋糕/餐具被拿起或放上早餐桌（机器人始终在厨房区域）；评分器 q_score=0、goal=0，与此一致。"
+     }
+    ]
+   },
+   "knives_next_to_plates_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程未见盘/蛋糕/餐具被拿起或放上早餐桌（机器人始终在厨房区域）；评分器 q_score=0、goal=0，与此一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：视频只见底盘与手臂摆动，未见夹爪抓住刀叉；日志 step 57 会话读数 grab k1 返回 None（会话证据，非评分回放）"
+     }
+    ]
+   },
+   "knives_next_to_plates_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 10.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程未见盘/蛋糕/餐具被拿起或放上早餐桌（机器人始终在厨房区域）；评分器 q_score=0、goal=0，与此一致。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "比例",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "knives_next_to_plates_1",
+   "t": 1.0,
+   "log_step": 57,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0-10.2 s 相机自约 0.5-1 s 起即翻滚倾斜，未见任何刀叉被拿起或放到早餐桌，也未见柜门打开；评分器 fallen=1。推断：交付脚本计划 A 段先做 knife1+fork1（log step 55 gen.py；step 57 会话输出首个动作为 k1 grab -> None），knife/fork 节点在 DAG 中无 after，故首个被尝试且失败的是刀叉放置（knives_next_to_plates_1）；柜门不是首个动作。此顺序来自脚本，视频无法直接确认。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 1.0,
+    "log_step": null,
+    "evidence_zh": "观察：评分器 fallen=1、sim_steps=305（1583 个动作只执行 305 步即终止）；视频时钟约 0.5-1 s 起画面已大幅倾斜翻转，约 3-10 s 画面基本稳定，推断为已倒状态，倒下时刻约在 1 s 前后，难以精确定位。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "轨迹或逆解",
+    "alt_feasible": "未知",
+    "evidence_zh": "区分：会话读数 vs 评分回放。会话读数（非评分回放）：log step 57/60 reset 后 cuRobo 无 IK 解、err 0.3-1.2、grab 返回 None（k1/f1），agent 自述为双臂 IK 无解。评分回放：1583 个动作仅 305 步即终止于 fallen=1；会话读数不能解释回放中为何倒下，二者关系未验证。故此根因为低置信推断。",
+    "source_zh": "日志 / agent 自述"
+   }
+  ],
+  "gap_zh": "无状态日志，头部相机画面翻滚；倒下时刻只能粗判（约 1 s）；根因来自 agent 会话读数而非评分回放，二者关系未知；断点依据脚本中 A 段顺序，视频不能直接确认。",
+  "unmet": {
+   "plates_on_table_1": false,
+   "plates_on_table_2": false,
+   "cupcakes_on_plates_1": false,
+   "cupcakes_on_plates_2": false,
+   "forks_next_to_plates_1": false,
+   "forks_next_to_plates_2": false,
+   "knives_next_to_plates_1": false,
+   "knives_next_to_plates_2": false
+  }
+ },
+ "summary_zh": "观察：10.2 s 的图传视频里机器人姿态很早失稳（约 1 s 起画面倾斜翻转，评分器 fallen=1），没有任何盘、蛋糕、餐具被移动，也未见柜门或冰箱门打开，0/10 节点达成（q_score 0）。断点为计划 A 段首步放置刀叉（knives_next_to_plates_1，操作中，依据脚本顺序）；agent 会话读数显示 reset 后双臂 IK 无解、grab 返回 None，推断为轨迹或逆解问题，但与回放倒下的关系未验证。主要缺口：无状态日志，视角混乱。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "断点 ref 选错：交付轨迹的计划 A 段先做刀叉（knife1/fork1，log step 57 首个动作 grab k1），DAG 中刀叉节点无 after、与 cabinet_opened 并列，所以首个失败步骤应是 knives_next_to_plates_1，而不是柜门。根因完全来自 agent 会话读数（reset 后 IK 无解），并未说明与评分回放中 fallen 的关系，且 phenomenon 时间与断点不一致。",
+    "n_issues": 4,
+    "n_major": 3
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 5,
+    "summary_zh": "标签与证据一致：视频 0-10.2 s 相机早期翻滚、评分器 fallen=1、q_score=0，0 节点达成；断点按脚本 A 段顺序选刀叉节点并如实标注为推断，根因标为低置信并区分会话与回放。未发现需改内容的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:56Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-setting-the-table-i00-privileged--pipwpla/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-setting-the-table-i00-privileged--pipwpla/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-setting-the-table-i00-privileged--pipwpla/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-setting-the-table-i00-privileged--pipwpla/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "20eba76fa40272c9cfd8ceb18e003a697de1995c6e65dd89db2af345a1b9cd93",
+   "video_sha256": "674118fa1e50038ea540c77cfaeaf5c64ad68604780c78c8ac2d06e82152386c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/sorting_books_on_shelf/trajectories/behavior-sorting-books-on-shelf-i00-privileged--pvddz3t.json b/phaseR/behavior-1k/sorting_books_on_shelf/trajectories/behavior-sorting-books-on-shelf-i00-privileged--pvddz3t.json
new file mode 100644
index 0000000000000000000000000000000000000000..96590ac599c10048ff910a5d8c8c0917ce61199c
--- /dev/null
+++ b/phaseR/behavior-1k/sorting_books_on_shelf/trajectories/behavior-sorting-books-on-shelf-i00-privileged--pvddz3t.json
@@ -0,0 +1,407 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-sorting-books-on-shelf-i00-privileged--pvddz3t",
+ "taskKey": "behavior-1k/sorting_books_on_shelf",
+ "dag_sha256": "a22c5cd43f8657b5acce376e54ea380253075b83b7eeb3679bc2dba6d608b7ce",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 0,
+   "n_actions": 2903,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 3053,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 797.8,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 5,
+   "agent_sim_steps": 0,
+   "agent_records": 2,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 8.33,
+   "peak_memory_gb": 12.649,
+   "peak_processes": 49,
+   "peak_disk_gb": 0.522,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-sorting-books-on-shelf-i00-privileged--pvddz3t.mp4",
+  "duration_s": 101.767,
+  "fps": 30.0,
+  "frames": 3053,
+  "sha256": "919dc5c7f02827822bef0923fc2a80da6bfa488b7e190ef649b5e03a065809a9"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 11,
+  "base": 7,
+  "peak": 7,
+  "final": 7,
+  "nodes": {
+   "comics_in_bookcase_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": 38,
+      "text_zh": "观察：agent 探索日志（log step 38，会话证据）称七本书初态均已在书柜内；视频 0-101.5 s 的帧中看不到书被拿出柜外或掉落。推断：入柜类节点自 t=0 成立并保持；视角多数时间看不到书，置信度有限。"
+     }
+    ]
+   },
+   "comics_in_bookcase_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": 38,
+      "text_zh": "观察：agent 探索日志（log step 38，会话证据）称七本书初态均已在书柜内；视频 0-101.5 s 的帧中看不到书被拿出柜外或掉落。推断：入柜类节点自 t=0 成立并保持；视角多数时间看不到书，置信度有限。"
+     }
+    ]
+   },
+   "comics_in_bookcase_3": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": 38,
+      "text_zh": "观察：agent 探索日志（log step 38，会话证据）称七本书初态均已在书柜内；视频 0-101.5 s 的帧中看不到书被拿出柜外或掉落。推断：入柜类节点自 t=0 成立并保持；视角多数时间看不到书，置信度有限。"
+     }
+    ]
+   },
+   "notebooks_in_bookcase_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": 38,
+      "text_zh": "观察：agent 探索日志（log step 38，会话证据）称七本书初态均已在书柜内；视频 0-101.5 s 的帧中看不到书被拿出柜外或掉落。推断：入柜类节点自 t=0 成立并保持；视角多数时间看不到书，置信度有限。"
+     }
+    ]
+   },
+   "notebooks_in_bookcase_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": 38,
+      "text_zh": "观察：agent 探索日志（log step 38，会话证据）称七本书初态均已在书柜内；视频 0-101.5 s 的帧中看不到书被拿出柜外或掉落。推断：入柜类节点自 t=0 成立并保持；视角多数时间看不到书，置信度有限。"
+     }
+    ]
+   },
+   "hardbacks_in_bookcase_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": 38,
+      "text_zh": "观察：agent 探索日志（log step 38，会话证据）称七本书初态均已在书柜内；视频 0-101.5 s 的帧中看不到书被拿出柜外或掉落。推断：入柜类节点自 t=0 成立并保持；视角多数时间看不到书，置信度有限。"
+     }
+    ]
+   },
+   "hardbacks_in_bookcase_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0,
+      "log_step": 38,
+      "text_zh": "观察：agent 探索日志（log step 38，会话证据）称七本书初态均已在书柜内；视频 0-101.5 s 的帧中看不到书被拿出柜外或掉落。推断：入柜类节点自 t=0 成立并保持；视角多数时间看不到书，置信度有限。"
+     }
+    ]
+   },
+   "comics_stacked_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 101.5,
+      "log_step": null,
+      "text_zh": "观察：视频 101.5 s 末帧头部/腕部视角只见书柜侧板与地板，没有任何书叠放完成的画面；全程未见书被抓起或移到另一本上。推断：叠放未达成（q_score=0 与此一致）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频视角几乎看不到书本，无法确认夹爪是否碰到书。"
+     }
+    ]
+   },
+   "comics_stacked_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 101.5,
+      "log_step": null,
+      "text_zh": "观察：视频 101.5 s 末帧头部/腕部视角只见书柜侧板与地板，没有任何书叠放完成的画面；全程未见书被抓起或移到另一本上。推断：叠放未达成（q_score=0 与此一致）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频视角几乎看不到书本，无法确认夹爪是否碰到书。"
+     }
+    ]
+   },
+   "notebooks_stacked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 101.5,
+      "log_step": null,
+      "text_zh": "观察：视频 101.5 s 末帧头部/腕部视角只见书柜侧板与地板，没有任何书叠放完成的画面；全程未见书被抓起或移到另一本上。推断：叠放未达成（q_score=0 与此一致）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频视角几乎看不到书本，无法确认夹爪是否碰到书。"
+     }
+    ]
+   },
+   "hardbacks_stacked": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 101.5,
+      "log_step": null,
+      "text_zh": "观察：视频 101.5 s 末帧头部/腕部视角只见书柜侧板与地板，没有任何书叠放完成的画面；全程未见书被抓起或移到另一本上。推断：叠放未达成（q_score=0 与此一致）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频视角几乎看不到书本，无法确认夹爪是否碰到书。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "hardbacks_stacked",
+   "where": "操作中",
+   "t": null,
+   "log_step": 86,
+   "evidence_zh": "推断：会话日志（非图像；交付的 trajectory.npz 为 13:34 第一次运行，对应 log step 86，同样先 MOVE hardback_2 且 GRASP FAILED；step 92 为第二次运行，未写出新 npz）显示抓取 in_hand=None。视频 0-101.5 s 只见机器人在书柜前移动，没有书被抓起或叠放，视频不能证实接触情况；q_score=0、goal=0。"
+  },
+  "phenomena": [],
+  "root_causes": [
+   {
+    "evidence_zh": "观察：log step 59-70 agent 自述两指夹爪(开口0.1 m)对书角/书边的捏取反复失败，书比夹爪宽，需要角点或悬边捏取；是否存在可行替代：未知。",
+    "source_zh": "agent 自述 / 日志",
+    "alt_feasible": "未知",
+    "cat": "控制",
+    "item": "抓取和接触设计"
+   }
+  ],
+  "unmet": {
+   "comics_in_bookcase_1": true,
+   "comics_in_bookcase_2": true,
+   "comics_in_bookcase_3": true,
+   "notebooks_in_bookcase_1": true,
+   "notebooks_in_bookcase_2": true,
+   "hardbacks_in_bookcase_1": true,
+   "hardbacks_in_bookcase_2": true,
+   "comics_stacked_1": false,
+   "comics_stacked_2": false,
+   "notebooks_stacked": false,
+   "hardbacks_stacked": false
+  },
+  "gap_zh": "视频是头部+腕部视角，书本几乎不可见，无状态日志，因此书是否被碰、被推出柜外、夹爪是否接触到书均无法由视频或状态判定；抓取失败仅见于agent会话日志（叙述性，非图像/状态证据），故现象层留空；入柜类节点依据初态与未见移出。"
+ },
+ "summary_zh": "观察：视频101.8 s中机器人在书柜前移动，臂在柜侧，从未见到书被抓起或叠放；七个入柜节点自初态成立（视频未见移出），四个叠放节点均未达成，q_score=0。推断：断点在精装书叠放，依据会话日志log step 86（交付轨迹对应的第一次运行）；视频无法证实接触，现象层留空，置信度有限。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 3,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 5,
+    "summary_zh": "七个入柜节点自 t=0 成立、四个叠放节点未达成、base=peak=final=7 与 q_score=0、视频一致，这部分我认可。问题在失败层：断点与现象“没建立接触”只来自 agent 会话日志，且引用的 log step 92 是未被交付的第二次运行；交付的 trajectory.npz 是 13:34 的第一次运行（log step 86）。视频帧看不到任何夹爪与书的接触。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "fix",
+    "conf": 5,
+    "summary_zh": "WHERE 层（7个入柜节点自0成立、4个叠放未达成、base=peak=final=7、与q_score=0一致）认可。问题：失败现象“没建立接触”仅来自agent会话日志，回放视频看不到夹爪与书的接触，属叙述性证据；summary_zh仍引用已被改为86的log step 92并把推断写成“观察”。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 3,
+    "kind": "confirm",
+    "verdict": "agree",
+    "summary_zh": "两个问题的接受均已在 prev/trajectory.json 中正确落实：现象层清空并在 gap_zh 说明，summary_zh 改为 log step 86 并标明推断。",
+    "n_issues": 0,
+    "n_checks": 2
+   }
+  ],
+  "open_zh": [],
+  "confirm": {
+   "verdict": "confirmed",
+   "summary_zh": "两个问题的接受均已在 prev/trajectory.json 中正确落实：现象层清空并在 gap_zh 说明，summary_zh 改为 log step 86 并标明推断。",
+   "checks_ok": "2/2",
+   "n_factual_errors": 0
+  }
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:38Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-books-on-shelf-i00-privileged--pvddz3t/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-books-on-shelf-i00-privileged--pvddz3t/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-books-on-shelf-i00-privileged--pvddz3t/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-books-on-shelf-i00-privileged--pvddz3t/04_review_r2"
+   },
+   {
+    "stage": "reconcile",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-books-on-shelf-i00-privileged--pvddz3t/05_reconcile_r2"
+   },
+   {
+    "stage": "confirm",
+    "round": 3,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-books-on-shelf-i00-privileged--pvddz3t/06_confirm_r3"
+   }
+  ],
+  "confirm_prompt": "prompts/06_confirm.md",
+  "confirm_prompt_sha256": "b9f56c095f7ea5e609e58bb21e076f565766c370333d71c7a3512cffb4d78267",
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "daf945a8eec54cc238a1729e706ad6a756ab0bda4b24b044069f86633ce3988c",
+   "video_sha256": "919dc5c7f02827822bef0923fc2a80da6bfa488b7e190ef649b5e03a065809a9",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/sorting_bottles_cans_and_paper/trajectories/behavior-sorting-bottles-cans-and-paper-i00-privileged--asfkhgb.json b/phaseR/behavior-1k/sorting_bottles_cans_and_paper/trajectories/behavior-sorting-bottles-cans-and-paper-i00-privileged--asfkhgb.json
new file mode 100644
index 0000000000000000000000000000000000000000..e07b815013c4ed29678ebed2c48e0036ba43838d
--- /dev/null
+++ b/phaseR/behavior-1k/sorting_bottles_cans_and_paper/trajectories/behavior-sorting-bottles-cans-and-paper-i00-privileged--asfkhgb.json
@@ -0,0 +1,277 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-sorting-bottles-cans-and-paper-i00-privileged--asfkhgb",
+ "taskKey": "behavior-1k/sorting_bottles_cans_and_paper",
+ "dag_sha256": "c6a17fb377bf88080da35329a4db6429ac0880e966001b02e8b92aa5a65071e4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 1216,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 1366,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 853.3,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 3,
+   "agent_resets": 13,
+   "agent_sim_steps": 0,
+   "agent_records": 3,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 10.357,
+   "peak_memory_gb": 22.213,
+   "peak_processes": 76,
+   "peak_disk_gb": 2.008,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-sorting-bottles-cans-and-paper-i00-privileged--asfkhgb.mp4",
+  "duration_s": 45.534,
+  "fps": 30.0,
+  "frames": 1366,
+  "sha256": "af898371e8df52d1e29265749e785db5ed4ad4806dbcefcd5164862fdd6fb1f6"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 6,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "wine_bottles_in_bucket_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 45.0,
+      "log_step": 103,
+      "text_zh": "观察：视频 33-45 s 头部相机里桶内始终为空，地面桶旁没有任何物品被放入；评分器 q_score=0、goal=0（评分器字段）。agent 最终自述轨迹“carries nothing”（log step 103）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频 14-28 s 左腕相机里夹爪夹住酒瓶并随底盘携带，约 28.5 s 起酒瓶脱手，29.6 s 已落在地面上远处；此后桶内始终为空。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "观察：底盘在厨房内驶向酒瓶、再驶向桶。"
+     }
+    ]
+   },
+   "wine_bottles_in_bucket_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 45.0,
+      "log_step": 103,
+      "text_zh": "观察：视频 33-45 s 头部相机里桶内始终为空，地面桶旁没有任何物品被放入；评分器 q_score=0、goal=0（评分器字段）。agent 最终自述轨迹“carries nothing”（log step 103）。"
+     }
+    ]
+   },
+   "soda_cans_in_bucket_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 45.0,
+      "log_step": 103,
+      "text_zh": "观察：视频 33-45 s 头部相机里桶内始终为空，地面桶旁没有任何物品被放入；评分器 q_score=0、goal=0（评分器字段）。agent 最终自述轨迹“carries nothing”（log step 103）。"
+     }
+    ]
+   },
+   "soda_cans_in_bucket_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 45.0,
+      "log_step": 103,
+      "text_zh": "观察：视频 33-45 s 头部相机里桶内始终为空，地面桶旁没有任何物品被放入；评分器 q_score=0、goal=0（评分器字段）。agent 最终自述轨迹“carries nothing”（log step 103）。"
+     }
+    ]
+   },
+   "newspaper_in_bucket": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 45.0,
+      "log_step": 103,
+      "text_zh": "观察：视频 33-45 s 头部相机里桶内始终为空，地面桶旁没有任何物品被放入；评分器 q_score=0、goal=0（评分器字段）。agent 最终自述轨迹“carries nothing”（log step 103）。"
+     }
+    ]
+   },
+   "magazine_in_bucket": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 45.0,
+      "log_step": 103,
+      "text_zh": "观察：视频 33-45 s 头部相机里桶内始终为空，地面桶旁没有任何物品被放入；评分器 q_score=0、goal=0（评分器字段）。agent 最终自述轨迹“carries nothing”（log step 103）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "比例",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "wine_bottles_in_bucket_1",
+   "t": 28.5,
+   "log_step": 103,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 14-28 s 机器人夹着酒瓶驶向桶，约 28.5 s 酒瓶在携带途中脱手落地，未到桶（桶内始终为空，33-45 s）。"
+  },
+  "unmet": {
+   "wine_bottles_in_bucket_1": false,
+   "wine_bottles_in_bucket_2": false,
+   "soda_cans_in_bucket_1": false,
+   "soda_cans_in_bucket_2": false,
+   "newspaper_in_bucket": false,
+   "magazine_in_bucket": false
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 28.5,
+    "log_step": 103,
+    "evidence_zh": "观察：视频 14-28 s 左腕相机酒瓶随夹爪被携带，28.5-29.6 s 酒瓶脱手掉到地面，未入桶（桶内始终为空）。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "alt_feasible": "未知",
+    "evidence_zh": "视频只显示携带约 14 s 后脱手，帧无法判断是夹持、轨迹还是反馈控制；归为控制仅为保守归类，子类型不明。",
+    "source_zh": "视频观察（原因不可由帧确定）"
+   }
+  ],
+  "gap_zh": "无状态日志；视频为头/腕相机，脱手原因帧无法确定；agent 自述（抓取失败/空放置）与帧不一致，不作为证据。"
+ },
+ "summary_zh": "观察：该轨迹（1216 步）夹起酒瓶并携带约 14 s 驶向桶，约 28.5 s 途中脱手落地，六件物品无一入桶（q_score 0）；6 个节点均未达成。断点在第1个酒瓶入桶（操作中），现象为接触·没保持住，根因仅能归为控制（子类型不明）。缺口：无状态日志，脱手原因帧无法确定。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点全 false 与 q_score=0 一致，无误。但“没建立接触”与视频矛盾：左腕相机 14–28 s 酒瓶在夹爪中并随底盘一起被携带，约 28.5–29.5 s 才脱手掉落、滑出视野，未到桶。应为 接触·没保持住，断点 t≈28.5 s；agent 自述的“空手放置/抓取全失败”不可当作被评分回合的事实，根因证据也只是自述。",
+    "n_issues": 4,
+    "n_major": 3
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "第二轮标签与视频一致：酒瓶被夹住携带约14 s（视频20 s左腕相机可见），29 s已脱手，六节点均未达成与q_score=0一致；断点、现象（没保持住）与保守根因可接受。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:01Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-bottles-cans-and-paper-i00-privileged--asfkhgb/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-bottles-cans-and-paper-i00-privileged--asfkhgb/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-bottles-cans-and-paper-i00-privileged--asfkhgb/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-bottles-cans-and-paper-i00-privileged--asfkhgb/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "42ee058174170601df02baab12ec4aa520311b8dcf45c318febd014389f8d5a7",
+   "video_sha256": "af898371e8df52d1e29265749e785db5ed4ad4806dbcefcd5164862fdd6fb1f6",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/sorting_vegetables/trajectories/behavior-sorting-vegetables-i00-privileged--rkjkqep.json b/phaseR/behavior-1k/sorting_vegetables/trajectories/behavior-sorting-vegetables-i00-privileged--rkjkqep.json
new file mode 100644
index 0000000000000000000000000000000000000000..395092debfaea494108d2bb46d403177b1eb2742
--- /dev/null
+++ b/phaseR/behavior-1k/sorting_vegetables/trajectories/behavior-sorting-vegetables-i00-privileged--rkjkqep.json
@@ -0,0 +1,323 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-sorting-vegetables-i00-privileged--rkjkqep",
+ "taskKey": "behavior-1k/sorting_vegetables",
+ "dag_sha256": "f223bd60cf10453bf9bc51bdfdffbdfba7e9614ddc649875f859654d88e96aee",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 4271,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 0,
+   "goal_first_step": -1,
+   "sim_steps": 2438,
+   "fallen": 1,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 1093.7,
+   "replay_timeout_s": 4047.3,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 28,
+   "agent_sim_steps": 0,
+   "agent_records": 4,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-sorting-vegetables-i00-privileged--rkjkqep.mp4",
+  "duration_s": 81.267,
+  "fps": 30.0,
+  "frames": 2438,
+  "sha256": "b58602427dd78d0e4155129573d4c178a83bd93db5a23f06c7c089171d94c4d3"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 13,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bok_choy_in_bowl_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频画面剧烈翻转，看不到夹爪是否夹持过小白菜；agent 会话中的抓取尝试不作证据。"
+     }
+    ]
+   },
+   "bok_choy_in_bowl_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "bok_choy_in_bowl_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "onions_in_bowl_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "onions_in_bowl_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "onions_in_bowl_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "leeks_in_bowl_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "leeks_in_bowl_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "broccoli_in_bowl_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "broccoli_in_bowl_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "corn_in_bowl_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "corn_in_bowl_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   },
+   "corn_in_bowl_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 81.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-81.3 s 的头部/腕部画面里没有任何一只搅拌碗内出现蔬菜；画面大部分时间天旋地转（天花板、侧躺的地板），视频 81.0 s 地板在画面侧边，机器人倾倒。评分器 q_score=0、goal=0（评分器字段）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "bok_choy_in_bowl_1",
+   "t": 2.0,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 约 1.5-5 s 起头部相机转向天花板、侧向地板，此后直至 81.3 s 画面持续旋转或侧躺，从未出现蔬菜篮或入碗的蔬菜；评分器 fallen=1、q_score=0。推断：机器人在去取第一棵小白菜的过程中倒地，之后没有任何一步被修复。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 2.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频约 2-81 s 视角翻转、地板在画面侧边（视频 81.0 s 仍侧躺）；评分器字段 fallen=1。"
+   }
+  ],
+  "root_causes": [],
+  "unmet": {
+   "bok_choy_in_bowl_1": false,
+   "bok_choy_in_bowl_2": false,
+   "bok_choy_in_bowl_3": false,
+   "onions_in_bowl_1": false,
+   "onions_in_bowl_2": false,
+   "onions_in_bowl_3": false,
+   "leeks_in_bowl_1": false,
+   "leeks_in_bowl_2": false,
+   "broccoli_in_bowl_1": false,
+   "broccoli_in_bowl_2": false,
+   "corn_in_bowl_1": false,
+   "corn_in_bowl_2": false,
+   "corn_in_bowl_3": false
+  },
+  "gap_zh": "无状态日志，摔倒的具体时刻和原因（底盘/手臂控制、导航发散）视频无法判定；agent 日志只有会话内多次 mv diverging 的导航发散记录（log step 100），不是受评轨迹的证据，故根因留空。"
+ },
+ "summary_zh": "观察：受评轨迹 81.3 s 视频中没有任何蔬菜入碗（q_score=0），13 个节点均未达成。推断：机器人在第一步（小白菜入碗）开始阶段就倒地（评分器 fallen=1，视频约 2 s 起画面翻转，81 s 仍侧躺），断点为第 1 棵小白菜入碗，现象为全身·平衡。缺口：无状态日志，摔倒原因与接触情况无法确认。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "视频 0-81 s 画面持续翻转/倾倒，从未见蔬菜入碗，q_score=0、fallen=1；13 节点全 false、断点首步、全身·平衡、根因留空并写明 gap 均合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:55:36Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-vegetables-i00-privileged--rkjkqep/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-sorting-vegetables-i00-privileged--rkjkqep/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "4bc3b98412c9c8fc937fdc808d982b8cd41715adbb32aa18bdf849746d94463d",
+   "video_sha256": "b58602427dd78d0e4155129573d4c178a83bd93db5a23f06c7c089171d94c4d3",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/store_batteries/trajectories/behavior-store-batteries-i00-privileged--2w49ei8.json b/phaseR/behavior-1k/store_batteries/trajectories/behavior-store-batteries-i00-privileged--2w49ei8.json
new file mode 100644
index 0000000000000000000000000000000000000000..298c965ad5d549a7120424c50b3b7f05b82173d7
--- /dev/null
+++ b/phaseR/behavior-1k/store_batteries/trajectories/behavior-store-batteries-i00-privileged--2w49ei8.json
@@ -0,0 +1,231 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-store-batteries-i00-privileged--2w49ei8",
+ "taskKey": "behavior-1k/store_batteries",
+ "dag_sha256": "84b6014bbaa8953a93cb6f1cd12059ba67bf4170c0810be551277e0d13859c5a",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 2993,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 3143,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 644.3,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 3,
+   "agent_resets": 2,
+   "agent_sim_steps": 6165,
+   "agent_records": 1,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-store-batteries-i00-privileged--2w49ei8.mp4",
+  "duration_s": 104.767,
+  "fps": 30.0,
+  "frames": 3143,
+  "sha256": "dbcaeea13f6166083a4dafb45a71e40aa6cfa98cafe1910037e230aa6cf7fc83"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 4,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "cabinet_drawer_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 30,
+      "log_step": null,
+      "text_zh": "观察：视频 5-45 s 机器人（头部相机视角）停在植物边桌/落地灯一侧，夹爪在桌面附近；全程未见任何柜抽屉面板外拉、内腔露出（对照 oracle 约 10-15 s 可见）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频 5-45 s 夹爪在边桌附近，无法确认是否接触抽屉把手；未见抽屉被拉出。"
+     }
+    ]
+   },
+   "batteries_in_cabinet_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 104,
+      "log_step": null,
+      "text_zh": "观察：视频 76-104.5 s 左腕相机始终可见夹爪持着蓝黑色电池，头部视角朝向厨房传递窗，末帧电池仍在手中、未见入柜；评分器 q_score=0、goal=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "视频约 60-104.5 s 左夹爪抓持一节电池（腕部相机可见），底盘移动携带；未入柜。"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "携带电池驱动移动。"
+     }
+    ]
+   },
+   "batteries_in_cabinet_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 104,
+      "log_step": null,
+      "text_zh": "观察：全程没有第二节电池入柜的画面；评分器 q_score=0。"
+     }
+    ]
+   },
+   "batteries_in_cabinet_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 104,
+      "log_step": null,
+      "text_zh": "观察：全程没有三节入柜的画面；评分器 goal=0，success=0。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "cabinet_drawer_opened",
+   "t": 45,
+   "where": "转场",
+   "log_step": null,
+   "evidence_zh": "观察：视频 3-50 s 机器人停在植物边桌一带，头视角无柜子，未靠近任何目标柜、未见抽屉接触；52-60 s 转向咖啡桌取电池，之后持电池走向厨房传递窗，末帧仍持电池，无电池入柜。推断：第一步（开抽屉）从未开始，机器人未到达柜操作位；现象/根因因状态日志缺失不可判。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "视频为头部+腕部拼接小画面，抽屉把手接触不可见，无状态日志；无法判定未开抽屉的具体现象（没建立接触/位姿不准）及根因，故现象与根因留空。agent 日志显示其交付的是 gen.py 生成的 2993 步 npz，agent_replays=0（未回放验证），但本文件不据此断言根因。",
+  "unmet": {
+   "batteries_in_cabinet_1": false,
+   "batteries_in_cabinet_2": false,
+   "batteries_in_cabinet_3": false
+  }
+ },
+ "summary_zh": "观察：被评分的 104.8 s 视频中机器人未打开任何柜抽屉，也没有电池入柜（q_score=0），4 个节点中 0 个达成。推断：卡在第一步开抽屉，末段持电池走向传递窗。缺口：视角限制，无法判定具体失败现象与根因。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 5,
+    "summary_zh": "节点值（0/4）与评分器一致，空现象/根因带 gap_zh 可接受；但断点 where=操作中 缺乏依据：视频 3-50 s 机器人一直停在植物边桌，从未到达目标柜、未做任何抽屉操作，之后转去取咖啡桌上的电池。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点 0/4 与评分器 q_score=0 一致；视频 5-45 s 机器人停在植物边桌，之后转向咖啡桌取电池并走向传递窗，末帧仍持电池，断点（转场）与空现象/根因加 gap_zh 可接受。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:45Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-store-batteries-i00-privileged--2w49ei8/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-store-batteries-i00-privileged--2w49ei8/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-store-batteries-i00-privileged--2w49ei8/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-store-batteries-i00-privileged--2w49ei8/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "c900858af1c48ffe3538bab7b910eb4bf76109bb5c910c622d434f9815b54e53",
+   "video_sha256": "dbcaeea13f6166083a4dafb45a71e40aa6cfa98cafe1910037e230aa6cf7fc83",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/storing_food/trajectories/behavior-storing-food-i00-privileged--wc4s8x3.json b/phaseR/behavior-1k/storing_food/trajectories/behavior-storing-food-i00-privileged--wc4s8x3.json
new file mode 100644
index 0000000000000000000000000000000000000000..5908f2a837579bd13e29c87bf07d76ac2a0ce5ca
--- /dev/null
+++ b/phaseR/behavior-1k/storing_food/trajectories/behavior-storing-food-i00-privileged--wc4s8x3.json
@@ -0,0 +1,307 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-storing-food-i00-privileged--wc4s8x3",
+ "taskKey": "behavior-1k/storing_food",
+ "dag_sha256": "fa03aca76ba4e06fedf3262aedd2698e264e22693c18f1870342eeb1e357b147",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "deterministic": 0,
+   "missing_trajectory": 1,
+   "grader_error": 0,
+   "infra_error": 0,
+   "agent_sessions": 2,
+   "agent_resets": 0,
+   "agent_sim_steps": 0,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.681,
+   "peak_memory_gb": 28.783,
+   "peak_processes": 45,
+   "peak_disk_gb": 1.507,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 1,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_memory_gb": 1,
+   "over_oom": 1,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "none",
+  "url": null,
+  "duration_s": null,
+  "fps": null,
+  "frames": null,
+  "sha256": null
+ },
+ "state": null,
+ "progress": {
+  "method": "none",
+  "denominator": 9,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "cabinet_opened": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：无视频、无状态日志；评分器 missing_trajectory=1、agent_sim_steps=0，未提交任何轨迹，评分器没有逐节点字段，无法判定本节点（agent 在 log step 99-101 仍在探索/IK 探测，未做任何交付动作）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未交付任何轨迹"
+     }
+    ]
+   },
+   "oatmeal_in_cabinet_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：无视频、无状态日志；评分器 missing_trajectory=1、agent_sim_steps=0，未提交任何轨迹，评分器没有逐节点字段，无法判定本节点（agent 在 log step 99-101 仍在探索/IK 探测，未做任何交付动作）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未交付任何轨迹"
+     }
+    ]
+   },
+   "oatmeal_in_cabinet_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：无视频、无状态日志；评分器 missing_trajectory=1、agent_sim_steps=0，未提交任何轨迹，评分器没有逐节点字段，无法判定本节点（agent 在 log step 99-101 仍在探索/IK 探测，未做任何交付动作）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未交付任何轨迹"
+     }
+    ]
+   },
+   "chips_in_cabinet_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：无视频、无状态日志；评分器 missing_trajectory=1、agent_sim_steps=0，未提交任何轨迹，评分器没有逐节点字段，无法判定本节点（agent 在 log step 99-101 仍在探索/IK 探测，未做任何交付动作）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未交付任何轨迹"
+     }
+    ]
+   },
+   "chips_in_cabinet_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：无视频、无状态日志；评分器 missing_trajectory=1、agent_sim_steps=0，未提交任何轨迹，评分器没有逐节点字段，无法判定本节点（agent 在 log step 99-101 仍在探索/IK 探测，未做任何交付动作）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未交付任何轨迹"
+     }
+    ]
+   },
+   "olive_oil_in_cabinet_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：无视频、无状态日志；评分器 missing_trajectory=1、agent_sim_steps=0，未提交任何轨迹，评分器没有逐节点字段，无法判定本节点（agent 在 log step 99-101 仍在探索/IK 探测，未做任何交付动作）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未交付任何轨迹"
+     }
+    ]
+   },
+   "olive_oil_in_cabinet_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：无视频、无状态日志；评分器 missing_trajectory=1、agent_sim_steps=0，未提交任何轨迹，评分器没有逐节点字段，无法判定本节点（agent 在 log step 99-101 仍在探索/IK 探测，未做任何交付动作）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未交付任何轨迹"
+     }
+    ]
+   },
+   "sugar_in_cabinet_1": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：无视频、无状态日志；评分器 missing_trajectory=1、agent_sim_steps=0，未提交任何轨迹，评分器没有逐节点字段，无法判定本节点（agent 在 log step 99-101 仍在探索/IK 探测，未做任何交付动作）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未交付任何轨迹"
+     }
+    ]
+   },
+   "sugar_in_cabinet_2": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": null,
+      "text_zh": "观察：无视频、无状态日志；评分器 missing_trajectory=1、agent_sim_steps=0，未提交任何轨迹，评分器没有逐节点字段，无法判定本节点（agent 在 log step 99-101 仍在探索/IK 探测，未做任何交付动作）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未交付任何轨迹"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "cabinet_opened",
+   "t": null,
+   "log_step": 101,
+   "where": "首步前",
+   "evidence_zh": "观察：agent 会话 7200 s 预算内（log step 75-82）仅用于等待仿真会话加载（SESSION_READY 用了 1720 s，期间内存超限被标记 violations memory_gb），之后在 log step 83-101 只做探索脚本和 IK 探测，最后一步仍是发送探测脚本；评分器 missing_trajectory=1、agent_sim_steps=0、agent_records=0，没有交付轨迹，第一步（开柜门）从未开始。"
+  },
+  "unmet": {
+   "oatmeal_in_cabinet_1": null,
+   "oatmeal_in_cabinet_2": null,
+   "chips_in_cabinet_1": null,
+   "chips_in_cabinet_2": null,
+   "olive_oil_in_cabinet_1": null,
+   "olive_oil_in_cabinet_2": null,
+   "sugar_in_cabinet_1": null,
+   "sugar_in_cabinet_2": null
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无视频、无状态日志、未交付轨迹：没有被评分的机器人动作，所有节点 intervals/final 未知。会话主要耗在仿真加载（约 1720 s）和脚本调试上，但耗时/预算用尽不作为现象或根因，且没有证据说明具体的规划或控制错误，故现象与根因留空。"
+ },
+ "summary_zh": "agent 在 7200 s 预算内只完成环境加载、探索和 IK 探测（log step 2-101），没有提交轨迹（missing_trajectory=1，agent_sim_steps=0），九个节点均无法判定（无视频、无状态日志）。断点为首步“打开橱柜”之前；没有证据支持具体现象或根因。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "无视频、无状态日志、未交付轨迹（missing_trajectory=1）；各节点 unknown、断点首步前、现象/根因留空并附 gap_zh 均符合规则，未发现需改动之处。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:54:56Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-storing-food-i00-privileged--wc4s8x3/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-storing-food-i00-privileged--wc4s8x3/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "55b75bf2e8fa347834f2347215f54b6d924bda501fb8ffe3edcd951c4df9a64f",
+   "video_sha256": null,
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; no local video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/sweeping_garage/trajectories/behavior-sweeping-garage-i00-privileged--ifvyd9o.json b/phaseR/behavior-1k/sweeping_garage/trajectories/behavior-sweeping-garage-i00-privileged--ifvyd9o.json
new file mode 100644
index 0000000000000000000000000000000000000000..cd8d5b4067a3c11e428457c1bb5ed3928d133f61
--- /dev/null
+++ b/phaseR/behavior-1k/sweeping_garage/trajectories/behavior-sweeping-garage-i00-privileged--ifvyd9o.json
@@ -0,0 +1,227 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-sweeping-garage-i00-privileged--ifvyd9o",
+ "taskKey": "behavior-1k/sweeping_garage",
+ "dag_sha256": "f3475aaaadbe01ae8e931c4ff7e60b637f0158bfa5997c7f098b427fd6767c9d",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 2640,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 2790,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 707.7,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 0,
+   "agent_resets": 0,
+   "agent_sim_steps": 0,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.878,
+   "peak_memory_gb": 13.276,
+   "peak_processes": 45,
+   "peak_disk_gb": 0.548,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-sweeping-garage-i00-privileged--ifvyd9o.mp4",
+  "duration_s": 93.0,
+  "fps": 30.0,
+  "frames": 2790,
+  "sha256": "be89ac10a342d222fcf2cbb6dadfed07eeb6e998a6de28f6de04ab2fa7e72b46"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "floor_clear_of_sand": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 92.9,
+      "log_step": 94,
+      "text_zh": "观察：视频约 8-92.9 s 机器人持扫帚绕车清扫，头部相机的地面未见大块残留，但微小粒子无法在画面分辨。日志 step 91/97 agent 自述保存的这次轨迹在会话内重放后仍剩沙子 2 粒（墙边 x≈-9.26 或 -1.9/-5.0 附近），未清完；评分器 goal=0、q_score=0。推断：沙子节点未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "推/扫",
+      "note_zh": "右/单侧夹爪抓住扫帚（辅助抓取）拖着扫帚头沿地板移动，粒子被扫帚的粒子清除体积删除；视频约 8-10 s 起扫帚在夹爪中，直到 92.9 s"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在车库内移动并拖动扫帚（含绕车和转向），覆盖地面"
+     }
+    ]
+   },
+   "floor_clear_of_dust": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 92.9,
+      "log_step": 94,
+      "text_zh": "观察：视频约 8-92.9 s 机器人持扫帚绕车清扫，头部相机的地面未见大块残留，但微小粒子无法在画面分辨。日志 step 91/97 agent 自述保存的这次轨迹在会话内重放后仍剩灰尘 2 粒（墙边 x≈-9.26 或 -1.9/-5.0 附近），未清完；评分器 goal=0、q_score=0。推断：灰尘节点未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "推/扫",
+      "note_zh": "右/单侧夹爪抓住扫帚（辅助抓取）拖着扫帚头沿地板移动，粒子被扫帚的粒子清除体积删除；视频约 8-10 s 起扫帚在夹爪中，直到 92.9 s"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘在车库内移动并拖动扫帚（含绕车和转向），覆盖地面"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "floor_clear_of_sand",
+   "t": 92.9,
+   "log_step": 97,
+   "where": "操作中",
+   "evidence_zh": "观察：评分器 goal=0、q_score=0；视频中机器人持扫帚清扫。推断（依 agent 会话自述，视频无法证实）：沙子和灰尘各剩2粒在墙边，扫帚路径没有覆盖这些边缘粒子。"
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 92.9,
+    "log_step": 97,
+    "evidence_zh": "推断：依 agent 会话自述（日志 step 91/97，残余粒子坐标沙 (-5.05,5.42)、(-9.26,3.19)，尘 (-1.89,4.21)、(-9.26,4.79)），边缘粒子未被扫到，故位姿不准；视频无法证实，该坐标来自会话内读数而非被评分的重放。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "alt_feasible": "未知",
+    "evidence_zh": "agent 在日志 step 97 自述靠墙的极端边缘粒子未能清除；会话中夹取扫帚需反复尝试（log step 62-79），说明扫帚贴墙角覆盖受限；无直接证据区分轨迹还是规划。",
+    "source_zh": "agent 自述"
+   }
+  ],
+  "gap_zh": "视频无法分辨微小残余粒子，残余数量来自 agent 日志（会话重放，评分器 goal=0）；无状态日志。 位姿不准现象仅为对 agent 自述的推断，非观察。",
+  "unmet": {
+   "floor_clear_of_sand": false,
+   "floor_clear_of_dust": false
+  }
+ },
+ "summary_zh": "观察：机器人（agent 脚本开环轨迹）抓住扫帚绕车库清扫整个 93 s 视频，评分器 goal=0，q=0，沙子与灰尘两节点均未达成。agent 自述仍剩沙2粒、尘2粒在墙边。证据缺口：视频看不清小粒子，无状态日志。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点 final=false、断点 ref、q=0 都有评分器支持；但证据里有两处把无法由视频支持的内容写成了观察：扫帚“3-92.9 s 始终在夹爪中”与“视频中扫帚路径未能到达边缘位置”。残余粒子数/坐标来自 agent 会话内读数（会话里反复 s.reset()），不是被评分的重放。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "两节点 final=false、峰值0、断点 floor_clear_of_sand 与评分器 goal=0/q_score=0 一致；视频 8 s 起扫帚在手、全程清扫，微小粒子无法在画面分辨，标签已把残余粒子与位姿不准明确标为依 agent 自述的推断并写入 gap_zh。根因停在“控制”并 alt_feasible=未知，符合分类规则。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:14Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-sweeping-garage-i00-privileged--ifvyd9o/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-sweeping-garage-i00-privileged--ifvyd9o/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-sweeping-garage-i00-privileged--ifvyd9o/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-sweeping-garage-i00-privileged--ifvyd9o/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "d7a3b2b37017601b07b986a1b41d1b343570953be540fb9bd236502752043b27",
+   "video_sha256": "be89ac10a342d222fcf2cbb6dadfed07eeb6e998a6de28f6de04ab2fa7e72b46",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/tidying_living_room/trajectories/behavior-tidying-living-room-i00-privileged--pjpewfn.json b/phaseR/behavior-1k/tidying_living_room/trajectories/behavior-tidying-living-room-i00-privileged--pjpewfn.json
new file mode 100644
index 0000000000000000000000000000000000000000..0489f1ca4344c087f75fd2b27a2dd79ff24ec74f
--- /dev/null
+++ b/phaseR/behavior-1k/tidying_living_room/trajectories/behavior-tidying-living-room-i00-privileged--pjpewfn.json
@@ -0,0 +1,243 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-tidying-living-room-i00-privileged--pjpewfn",
+ "taskKey": "behavior-1k/tidying_living_room",
+ "dag_sha256": "6ad898fcc81887c7be8a7512909d9f7335f13eac54108e7edf40479563618d3a",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 937,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 1087,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 435.4,
+   "replay_timeout_s": 3600.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 8,
+   "agent_sim_steps": 0,
+   "agent_records": 2,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 8.514,
+   "peak_memory_gb": 12.656,
+   "peak_processes": 48,
+   "peak_disk_gb": 0.519,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-tidying-living-room-i00-privileged--pjpewfn.mp4",
+  "duration_s": 36.234,
+  "fps": 30.0,
+  "frames": 1087,
+  "sha256": "ce8c0bc85d458aae9b27d22f369c44d2a0099c634f4490f8333200191f8204e0"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 4,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "plant_on_table": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 35.0,
+      "log_step": null,
+      "text_zh": "观察：全程 36 s 视频只见机器人抓取蓝绿色精装书并搬向书柜，未见盆栽、报纸、笔记本被触碰或移动；末帧该物未处于目标关系。q_score=0（评分器字段）。"
+     }
+    ]
+   },
+   "newspaper_on_table": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 35.0,
+      "log_step": null,
+      "text_zh": "观察：全程 36 s 视频只见机器人抓取蓝绿色精装书并搬向书柜，未见盆栽、报纸、笔记本被触碰或移动；末帧该物未处于目标关系。q_score=0（评分器字段）。"
+     }
+    ]
+   },
+   "hardback_in_bookcase": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 30.0,
+      "log_step": null,
+      "text_zh": "观察：视频 ≈14-22 s 机器人夹持精装书，≈22-27 s 书仍被夹持并贴近书柜下缘；≈27-28 s 脱手落到书柜外侧地板，至 36.2 s 未进入格口，书中心不在柜内。评分器 q_score=0，亦不支持已达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "机器人一侧夹爪夹住精装书并搬运至书柜处（视频 ≈14-27 s）"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘驶向书柜期间承重并移动"
+     }
+    ]
+   },
+   "notebook_on_desk": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 35.0,
+      "log_step": null,
+      "text_zh": "观察：全程 36 s 视频只见机器人抓取蓝绿色精装书并搬向书柜，未见盆栽、报纸、笔记本被触碰或移动；末帧该物未处于目标关系。q_score=0（评分器字段）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": 0.0,
+   "unit": "1",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hardback_in_bookcase",
+   "t": 25.0,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 ≈22-27 s 书仍被夹持并贴近书柜下缘，≈27-28 s 脱手落到柜外地板，至 36.2 s 未进入格口；其余三项从未尝试。",
+   "log_step": null
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 27.5,
+    "log_step": null,
+    "evidence_zh": "观察：视频 ≈27 s 书仍在夹爪中、位于书柜口外下缘，28 s 夹爪已空、书躺在柜外地板，28-36 s 未进入格口（松手时书不在目标内）。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "无状态日志，无法区分根因（插入偏差、被柜挡住或感知误差），根因证据不足故留空；现象（位姿不准）为画面可见。agent 自述（log step 127）称插入经射线探测的空隙，但无视频佐证。",
+  "unmet": {
+   "plant_on_table": false,
+   "newspaper_on_table": false,
+   "hardback_in_bookcase": false,
+   "notebook_on_desk": false
+  }
+ },
+ "summary_zh": "观察：视频 36.2 s 内机器人仅抓取精装书并搬向书柜，≈27-28 s 在柜口外松手落地，未能放入，其余三项未尝试；q_score=0，0/4 节点达成。断点为精装书入书柜（操作中），现象位姿不准；根因因无状态日志留空。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点值（全 false、q_score=0）和断点 ref 正确；但视频可见书在 ≈27-28 s 才脱手落在书柜口外的地板上，label 写 ≈25 s 起书已在地板（偏早 2-3 s），且现象层留空——书在进入格口前松手是画面可见的现象（位姿不准），应补上。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点全 false、断点 hardback_in_bookcase、位姿不准现象与视频一致（书在柜外松手落地，未进格口），根因留空并附 gap_zh 合理；第一轮问题已修正。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:47Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-tidying-living-room-i00-privileged--pjpewfn/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-tidying-living-room-i00-privileged--pjpewfn/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-tidying-living-room-i00-privileged--pjpewfn/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-tidying-living-room-i00-privileged--pjpewfn/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3e7baff7842c8b543235d3de1d6feb353392f5a7089f8851ec9fc13b900d1d53",
+   "video_sha256": "ce8c0bc85d458aae9b27d22f369c44d2a0099c634f4490f8333200191f8204e0",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/unloading_the_car/trajectories/behavior-unloading-the-car-i00-privileged--jcbzbd3.json b/phaseR/behavior-1k/unloading_the_car/trajectories/behavior-unloading-the-car-i00-privileged--jcbzbd3.json
new file mode 100644
index 0000000000000000000000000000000000000000..8996b8255e3f1df639245a2a1c35c4194dda1417
--- /dev/null
+++ b/phaseR/behavior-1k/unloading_the_car/trajectories/behavior-unloading-the-car-i00-privileged--jcbzbd3.json
@@ -0,0 +1,260 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-unloading-the-car-i00-privileged--jcbzbd3",
+ "taskKey": "behavior-1k/unloading_the_car",
+ "dag_sha256": "2d43d8f6f9e62fd26631473e35e040c8d2125523130e79643cca5c9d50acd36d",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "all_success": 0,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 4928,
+   "sim_breakdown": 0,
+   "goal": 0,
+   "settle_steps": 150,
+   "goal_first_step": -1,
+   "sim_steps": 5078,
+   "fallen": 0,
+   "final_reward": 0.0,
+   "q_score": 0.0,
+   "replay_wall_s": 1677.6,
+   "replay_timeout_s": 4485.3,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 1,
+   "agent_sim_steps": 0,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.438,
+   "peak_memory_gb": 12.677,
+   "peak_processes": 47,
+   "peak_disk_gb": 1.488,
+   "peak_cpus": 8.04,
+   "peak_pids": 379,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-unloading-the-car-i00-privileged--jcbzbd3.mp4",
+  "duration_s": 169.267,
+  "fps": 30.0,
+  "frames": 5078,
+  "sha256": "104843962b7f7fad46701d7d66808667c30f57f6ca460d20e8a1879635bccb56"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 4,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "car_opened": {
+    "method": "visual",
+    "intervals": [
+     [
+      52.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 55.0,
+      "log_step": null,
+      "text_zh": "观察：视频 55 s 头部相机可见后备厢盖已抬起，开口足以露出车内；视频 44 s 时盖仍遮挡。时间约 ±5 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：双臂夹爪在视频约 20-55 s 作用于后备厢盖；具体接触方式据腕部相机推断"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "轮式底盘移动到车尾"
+     }
+    ]
+   },
+   "garage_exit_opened": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 168.0,
+      "log_step": null,
+      "text_zh": "观察：视频 160-169 s 头部相机见走廊门半开（门扇与门框间露出门后地面，开度约三分之一），未见机器人开门动作。推断：开度不足以携包通过，机器人未通过，q_score 0，故 final=false；开度是否达到判定阈值无日志可核，只能由画面推断。"
+     }
+    ]
+   },
+   "briefcase_next_to_sofa": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 168.0,
+      "log_step": null,
+      "text_zh": "观察：视频 168 s 包躺在车库地面，沙发所在客厅从未出现；q_score 0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：视频约 100-158 s 右夹爪夹着公文包，约 159 s 脱手"
+     },
+     {
+      "who": "底盘",
+      "how": "支撑",
+      "note_zh": "底盘携包移动"
+     }
+    ]
+   },
+   "satchel_next_to_sofa": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 168.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程未见挎包被带到沙发旁；q_score 0。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "briefcase_next_to_sofa",
+   "t": 159.0,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 100-158.5 s 右夹爪持包，约 159 s 包脱手落到车库地面，160-169 s 留在原地；走廊门半开但机器人未携包通过，视频结束时包未到沙发。",
+   "log_step": null
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 159.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频 155-158.5 s 右夹爪持包向走廊门方向行进，159-160 s 包落到车库地面并此后不动（搬运途中脱落）。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "没有状态日志，脱手的原因（抓取不稳还是控制）无法由视频区分，根因留空；agent 自述（log step 381）时间用完，未验证回放。",
+  "unmet": {
+   "briefcase_next_to_sofa": false,
+   "satchel_next_to_sofa": false
+  }
+ },
+ "summary_zh": "观察：图为 gpt-6-astra 的 privileged 运行。机器人在视频约 52 s 打开后备厢，抓起公文包搬运至约 158.5 s，约 159 s 包脱手落到车库地面，走廊门半开但机器人未通过，沙发旁无包，q_score 0。推断：断点为公文包搬运，现象为没保持住；根因无法由视频判定。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "断点（公文包搬运）与 car_opened 大体成立，但 (1) 车库门 168 s 实际已半开而文件写“门关闭、全程未见被打开”；(2) 视频清楚显示夹爪持包至约 158.5 s、约 159 s 脱手落地，却因“无状态日志”把现象留空，应标没保持住；(3) 持包区间写 100-140 s，实际到约 158 s。",
+    "n_issues": 3,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "后备厢约 46-52 s 打开、公文包持至约 158.5 s、159 s 脱手落在车库地面、门半开、沙发旁无包，均与帧一致；断点与现象（没保持住）成立，根因留空有 gap_zh。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:19Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-unloading-the-car-i00-privileged--jcbzbd3/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-unloading-the-car-i00-privileged--jcbzbd3/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-unloading-the-car-i00-privileged--jcbzbd3/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/behavior-unloading-the-car-i00-privileged--jcbzbd3/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "867807e97d4ef9c98b5a9fe711dc4243284ffa0d842e5478591c06501cf8b771",
+   "video_sha256": "104843962b7f7fad46701d7d66808667c30f57f6ca460d20e8a1879635bccb56",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/wash_a_baseball_cap/trajectories/behavior-wash-a-baseball-cap-i00-privileged--moflayy.json b/phaseR/behavior-1k/wash_a_baseball_cap/trajectories/behavior-wash-a-baseball-cap-i00-privileged--moflayy.json
new file mode 100644
index 0000000000000000000000000000000000000000..a22f86d005d9d6c409944465b22cc003c367e0cb
--- /dev/null
+++ b/phaseR/behavior-1k/wash_a_baseball_cap/trajectories/behavior-wash-a-baseball-cap-i00-privileged--moflayy.json
@@ -0,0 +1,160 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-wash-a-baseball-cap-i00-privileged--moflayy",
+ "taskKey": "behavior-1k/wash_a_baseball_cap",
+ "dag_sha256": "053fa67e538181953dbdeadf3cc6214fb48ffaf32dc990187116a191052c4c85",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "deterministic": 0,
+   "missing_trajectory": 1,
+   "grader_error": 0,
+   "infra_error": 0,
+   "agent_sessions": 1,
+   "agent_resets": 38,
+   "agent_sim_steps": 0,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 9.995,
+   "peak_memory_gb": 12.801,
+   "peak_processes": 47,
+   "peak_disk_gb": 0.807,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "none",
+  "url": null,
+  "duration_s": null,
+  "fps": null,
+  "frames": null,
+  "sha256": null
+ },
+ "state": null,
+ "progress": {
+  "method": "log",
+  "denominator": 2,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "caps_clean_1": {
+    "method": "log",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 138,
+      "text_zh": "观察：无视频、无状态日志；reward 记录 missing_trajectory=1、agent_records=0、agent_replays=0，没有交付可回放轨迹，评分 success=0。agent 的会话内实验（log step 133-138）是会话证据，不作节点值。"
+     }
+    ]
+   },
+   "caps_clean_2": {
+    "method": "log",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": null,
+      "log_step": 138,
+      "text_zh": "观察：无视频、无状态日志；reward 记录 missing_trajectory=1、agent_records=0、agent_replays=0，没有交付可回放轨迹，评分 success=0。agent 的会话内实验（log step 133-138）是会话证据，不作节点值。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "q_score",
+   "value": null,
+   "unit": "1",
+   "threshold": 1.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "caps_clean_1",
+   "where": "首步前",
+   "t": null,
+   "log_step": 138,
+   "evidence_zh": "观察：run.json reward 显示 missing_trajectory=1、agent_records=0、agent_replays=0；agent 在约 7200 s 预算内（最后 log step 138，agent 工作约 7011 s）一直在做会话内实验（如 log step 133-137 测试夹爪能否把帽子伸入洗衣机滚筒），没有交付轨迹。无视频，故未见到被评分的执行。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无视频、无状态日志，且没有交付轨迹（missing_trajectory=1）；会话实验只是会话证据，不能作为现象或根因。agent 因自身原因未交付，故现象与根因留空。q_score 未在 reward 字段中报告。",
+  "unmet": {
+   "caps_clean_1": null,
+   "caps_clean_2": null
+  }
+ },
+ "summary_zh": "观察：未交付任何轨迹（missing_trajectory=1，success=0），无视频与状态日志，两顶帽子节点均无法按视频判定，final 记为未知。agent 在约 7200 s 预算内反复实验把帽子伸入洗衣机滚筒（log step 133-137 自述伸展受阻），始终未提交。断点记为第1顶帽洗净、首步前；现象与根因因证据不足留空。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "无视频、无状态日志、未交付轨迹（missing_trajectory=1）；节点 final 均为 null、现象与根因留空并写 gap_zh、断点首步前，符合 N9 与 D23/FAIL-22。时间均标明为 agent 工作时间，未转换成视频时间。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:55:22Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-wash-a-baseball-cap-i00-privileged--moflayy/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-wash-a-baseball-cap-i00-privileged--moflayy/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "752ed0ee7a8312da4528c780bed8a09674b514b02fdc4277b340c68c9463ea9a",
+   "video_sha256": null,
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; no local video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/behavior-1k/wash_dog_toys/trajectories/behavior-wash-dog-toys-i00-privileged--shno6fv.json b/phaseR/behavior-1k/wash_dog_toys/trajectories/behavior-wash-dog-toys-i00-privileged--shno6fv.json
new file mode 100644
index 0000000000000000000000000000000000000000..cf437cfbb0f6ab3061eb94bcaab025174c40c92b
--- /dev/null
+++ b/phaseR/behavior-1k/wash_dog_toys/trajectories/behavior-wash-dog-toys-i00-privileged--shno6fv.json
@@ -0,0 +1,222 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "behavior-wash-dog-toys-i00-privileged--shno6fv",
+ "taskKey": "behavior-1k/wash_dog_toys",
+ "dag_sha256": "30841e135d3b6dc1597358bd99c28596bf89dbe6d7f5d69955eb8b7e69f236fe",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "all_success": 1,
+   "deterministic": 1,
+   "instance_ok": 1,
+   "infra_error": 0,
+   "instance_exact": 1,
+   "n_actions": 25076,
+   "sim_breakdown": 0,
+   "goal": 1,
+   "settle_steps": 0,
+   "goal_first_step": 25047,
+   "sim_steps": 25076,
+   "fallen": 0,
+   "final_reward": 1.0,
+   "q_score": 1.0,
+   "replay_wall_s": 4256.3,
+   "replay_timeout_s": 17917.3,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "video_replay_matches": 1,
+   "agent_sessions": 1,
+   "agent_resets": 0,
+   "agent_sim_steps": 0,
+   "agent_records": 0,
+   "agent_replays": 0,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:behavior-wash-dog-toys-i00-privileged--shno6fv.mp4",
+  "duration_s": 835.867,
+  "fps": 30.0,
+  "frames": 25076,
+  "sha256": "5f89aeb7c951fd05bb3d88275d190f74934508aef7e4f671b847be95b37c0010"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 4,
+  "base": 0,
+  "peak": 4,
+  "final": 4,
+  "nodes": {
+   "teddies_clean_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      834.9,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 834.9,
+      "log_step": 362,
+      "text_zh": "观察：评分器字段 goal_first_step=25047（30 fps → 视频 834.9 s）、success=1、q_score=1.0，四件玩具均干净、离手静止；视频 780-830 s 可见垒球被夹爪送入洗衣机、滚筒内可见熊与球，机器人随后关门按开关。推断：清洁由洗衣机完成，四件玩具同时在最后一次洗涤后才全部满足，故各节点按总成功时刻 834.9 s 记为达成（清洁具体起点无状态日志，可能略早）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "无接触",
+      "how": "不适用",
+      "note_zh": "洗衣机完成清洁；机器人双臂夹爪此前抓取玩具装入洗衣机、关门并按开关（视频 780-830 s 可见垒球装入）。"
+     }
+    ]
+   },
+   "teddies_clean_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      834.9,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 834.9,
+      "log_step": 362,
+      "text_zh": "观察：评分器字段 goal_first_step=25047（30 fps → 视频 834.9 s）、success=1、q_score=1.0，四件玩具均干净、离手静止；视频 780-830 s 可见垒球被夹爪送入洗衣机、滚筒内可见熊与球，机器人随后关门按开关。推断：清洁由洗衣机完成，四件玩具同时在最后一次洗涤后才全部满足，故各节点按总成功时刻 834.9 s 记为达成（清洁具体起点无状态日志，可能略早）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "无接触",
+      "how": "不适用",
+      "note_zh": "洗衣机完成清洁；机器人双臂夹爪此前抓取玩具装入洗衣机、关门并按开关（视频 780-830 s 可见垒球装入）。"
+     }
+    ]
+   },
+   "tennis_ball_clean": {
+    "method": "visual",
+    "intervals": [
+     [
+      834.9,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 834.9,
+      "log_step": 362,
+      "text_zh": "观察：评分器字段 goal_first_step=25047（30 fps → 视频 834.9 s）、success=1、q_score=1.0，四件玩具均干净、离手静止；视频 780-830 s 可见垒球被夹爪送入洗衣机、滚筒内可见熊与球，机器人随后关门按开关。推断：清洁由洗衣机完成，四件玩具同时在最后一次洗涤后才全部满足，故各节点按总成功时刻 834.9 s 记为达成（清洁具体起点无状态日志，可能略早）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "无接触",
+      "how": "不适用",
+      "note_zh": "洗衣机完成清洁；机器人双臂夹爪此前抓取玩具装入洗衣机、关门并按开关（视频 780-830 s 可见垒球装入）。"
+     }
+    ]
+   },
+   "softball_clean": {
+    "method": "visual",
+    "intervals": [
+     [
+      834.9,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 834.9,
+      "log_step": 362,
+      "text_zh": "观察：评分器字段 goal_first_step=25047（30 fps → 视频 834.9 s）、success=1、q_score=1.0，四件玩具均干净、离手静止；视频 780-830 s 可见垒球被夹爪送入洗衣机、滚筒内可见熊与球，机器人随后关门按开关。推断：清洁由洗衣机完成，四件玩具同时在最后一次洗涤后才全部满足，故各节点按总成功时刻 834.9 s 记为达成（清洁具体起点无状态日志，可能略早）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "无接触",
+      "how": "不适用",
+      "note_zh": "洗衣机完成清洁；机器人双臂夹爪此前抓取玩具装入洗衣机、关门并按开关（视频 780-830 s 可见垒球装入）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": null,
+ "summary_zh": "成功：特权模式下机器人用夹爪把两只熊、网球、垒球逐个装入洗衣机，关门并按开关；评分器 goal_first_step=25047（视频 834.9 s）四件玩具均干净、离手静止。无状态日志，各节点达成时间按总成功时刻给出，清洁具体起点未能单独核对。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "成功运行，评分器 goal_first_step=25047（834.9 s）、q_score=1.0，四个节点按此记为达成且已声明为推断；未发现需要修改的错误。各节点的清洁时刻可能略早，但没有状态日志可核对，标注已如实说明。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:17Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/behavior-wash-dog-toys-i00-privileged--shno6fv/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/behavior-wash-dog-toys-i00-privileged--shno6fv/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "a94140b95b56f26b4ace2ff21533017d50bda8197345251c42860b8beb43f0bd",
+   "video_sha256": "5f89aeb7c951fd05bb3d88275d190f74934508aef7e4f671b847be95b37c0010",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/backflip/trajectories/embodied-olympics-fun-backflip-i00-standard--vdq3an5.json b/phaseR/embodied-olympics-fun/backflip/trajectories/embodied-olympics-fun-backflip-i00-standard--vdq3an5.json
new file mode 100644
index 0000000000000000000000000000000000000000..dcbca05e8c0b693fae505a3ac4ab52ee53243ba4
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/backflip/trajectories/embodied-olympics-fun-backflip-i00-standard--vdq3an5.json
@@ -0,0 +1,213 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-backflip-i00-standard--vdq3an5",
+ "taskKey": "embodied-olympics-fun/backflip",
+ "dag_sha256": "a2dce89dd7f0391d67651e778f82fd48e6fb3121b94be3c140496cb25fc4241d",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 118,
+   "sim_steps": 414,
+   "calls": 9,
+   "rows": 414,
+   "resets": 2,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 6903.7,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 3.0,
+   "peak_processes": 15,
+   "peak_disk_gb": 0.326,
+   "peak_cpus": 5.67,
+   "peak_pids": 365,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_cpus": 1
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-backflip-i00-standard--vdq3an5.mp4",
+  "duration_s": 1.2,
+  "fps": 25.0,
+  "frames": 30,
+  "sha256": "993633d3c9b54726d1083473102ce356653aa171dca11c120c99f9f8a155d02c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "backflip_landed_stable": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.1,
+      "log_step": 221,
+      "text_zh": "观察：视频约 0.3-0.9 s 机器人起跳并在侧视中后翻，约 1.0-1.2 s 落地时躯干前倾、膝弯曲、手臂前伸，未直身站稳，视频在落地后不久结束（1.2 s）。log step 221 服务端读数 outcome=fell、stages_reached 3/4、sim_time 1.18 s：站稳验收未完成；各阶段含义文件未说明，翻腾是否达标未知。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：双脚蹬地起跳，落地时脚先着地但随后失稳前倾，视频末帧身体前倾接近地面。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "turn_deg",
+   "value": null,
+   "unit": "deg",
+   "threshold": 300.0,
+   "direction": "higher_better"
+  },
+  {
+   "name": "air_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 0.2,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "backflip_landed_stable",
+   "t": 1.1,
+   "log_step": 221,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 ≈1.0-1.2 s 落地后躯干前倾、未直身，视频结束；log step 221 服务读数 outcome=fell，stages 3/4（仿真 1.18 s）；log step 222 服务 proprio 骨盆俯仰 1.0-1.1 s 约 +66°（折算后转角约 294°）。翻腾转角是否达 300°、哪一阶段未过，无法由 stages 3/4 判定，未知。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 1.1,
+    "log_step": 221,
+    "evidence_zh": "观察：落地后身体前倾、手臂前伸，log step 221 判定 outcome=fell；step 222 服务 proprio 落地时骨盆俯仰约 +66°。平衡归类按落地失稳的可见表现；转角不足（起跳/翻转轨迹问题）的可能性未排除。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "无状态日志，回放仅 1.2 s，落地瞬间细节看不清；stages 3/4 含义未说明，落地时折算转角约 294° 低于 300°，翻腾是否达标与失败是落地平衡还是转角不足无法区分；根因无证据，留空。",
+  "unmet": {
+   "backflip_landed_stable": false
+  }
+ },
+ "summary_zh": "观察：机器人起跳后翻，服务读数 3/4 阶段（log step 221），落地后前倾失稳判定 fell，未完成站稳验收。落地俯仰约 +66°（折算转角约 294°，step 222），翻腾是否达标未知。断点在落地站稳；根因证据不足。无状态日志，转角未直接测得。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点 backflip_landed_stable=false、断点位置、时间（视频≈1.0-1.2 s，仿真 1.18 s）与 log step 221 的 outcome=fell、stages 3/4 一致。但“翻腾已完成/落地站稳阶段失败”写成推断依据不足：log step 222 的服务端 proprio（graded episode 第3次）显示骨盆俯仰 1.0-1.1 s 为 +66.4°（折算后向转角≈294°，低于300°），落地时躯干前倾约 66°，stages 3/4 的含义文件未说明。平衡现象与根因留空的处理需要相应收紧。",
+    "n_issues": 2,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点 false、断点在落地站稳、时间（视频≈1.0-1.2 s）与 log step 221 服务读数（outcome=fell，stages 3/4，sim 1.18 s）一致；推断部分已标明未知，平衡现象的不确定性在 gap_zh 中如实说明，根因留空有解释。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:12Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-backflip-i00-standard--vdq3an5/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-backflip-i00-standard--vdq3an5/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-backflip-i00-standard--vdq3an5/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-backflip-i00-standard--vdq3an5/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "dbbd6fa3060254fc91e0843441b7a0ea636044daffe82f1ed6b81abedfd46b3f",
+   "video_sha256": "993633d3c9b54726d1083473102ce356653aa171dca11c120c99f9f8a155d02c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/badminton_serve/trajectories/embodied-olympics-fun-badminton-serve-i00-standard--5wwjcun.json b/phaseR/embodied-olympics-fun/badminton_serve/trajectories/embodied-olympics-fun-badminton-serve-i00-standard--5wwjcun.json
new file mode 100644
index 0000000000000000000000000000000000000000..f8d8186b5a5375e037702df8f86e5c7ca3b4ee6d
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/badminton_serve/trajectories/embodied-olympics-fun-badminton-serve-i00-standard--5wwjcun.json
@@ -0,0 +1,167 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-badminton-serve-i00-standard--5wwjcun",
+ "taskKey": "embodied-olympics-fun/badminton_serve",
+ "dag_sha256": "e5772140aff5f054a271f90fd295419869c755e083b6526601e08a99bee573e1",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "calls": 1,
+   "rows": 0,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 3510.6,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 4,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.49,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.322,
+   "peak_cpus": 4.01,
+   "peak_pids": 175,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-badminton-serve-i00-standard--5wwjcun.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "7f28443388c68edd77037540350bf291bc019550e5b1cbe4705cd6e929f008a5"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "serve_in_target": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频只有 1 帧（0.04 s），显示机器人站在近侧发球区、球拍在手，尚无放球或击球；run.json 中 n_actions=0、rows=0、calls=1，没有交付任何轨迹。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付轨迹，机器人未执行任何动作。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "unknown",
+   "t": null,
+   "log_step": 129,
+   "where": "首步前",
+   "evidence_zh": "观察：log step 129、131、133、135 的回复均为 Gemini 过滤器拦截提示（“This request was blocked by Gemini's filters”），无任何动作调用；此前 agent 只在本地 MuJoCo 中调参（log step 122-128），run.json n_actions=0、rows=0。视频仅 1 帧，机器人未动。推断：agent 在交付任何轨迹之前被模型提供方的过滤器截断，属外部因素。"
+  },
+  "unmet": {
+   "serve_in_target": false
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "外部因素：模型提供方（Gemini）的过滤器在 log step 129 起截断了 agent，且未交付任何轨迹（n_actions=0），按规则不给现象和根因；视频仅 1 帧，无法观察机器人行为。"
+ },
+ "summary_zh": "观察：视频仅 1 帧，agent 未交付任何轨迹（n_actions=0）；log step 129 起被 Gemini 过滤器拦截，此前只在本地仿真里调参。唯一节点“合法发球入区”未达成。推断：这是提供方过滤器造成的外部因素截断，不属于机器人失败，不给现象与根因。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标注正确：run.json 全程 submit=0、n_actions=0，最后 4 步均为 Gemini 过滤器拦截且无调用，属外部因素截断、未交付任何轨迹，现象/根因为空并有 gap_zh；唯一节点未达成，视频仅 1 帧，时间轴无误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:55:53Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-badminton-serve-i00-standard--5wwjcun/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-badminton-serve-i00-standard--5wwjcun/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "79e9e10ec61610f885db430220e42fce222109ed9af5052bc319193842c503e8",
+   "video_sha256": "7f28443388c68edd77037540350bf291bc019550e5b1cbe4705cd6e929f008a5",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/balance_beam/trajectories/embodied-olympics-fun-balance-beam-i00-privileged--fj5xcvd.json b/phaseR/embodied-olympics-fun/balance_beam/trajectories/embodied-olympics-fun-balance-beam-i00-privileged--fj5xcvd.json
new file mode 100644
index 0000000000000000000000000000000000000000..5a47c23f23fa0abfee2f3417215774b591fb850b
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/balance_beam/trajectories/embodied-olympics-fun-balance-beam-i00-privileged--fj5xcvd.json
@@ -0,0 +1,162 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-balance-beam-i00-privileged--fj5xcvd",
+ "taskKey": "embodied-olympics-fun/balance_beam",
+ "dag_sha256": "85f86504f4e58cb42766772925fcd26453cfbc2255676f62e21bacd7398ab6ac",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 7189,
+   "sim_steps": 7189,
+   "horizon": 9000,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 3,
+   "stages_total": 3,
+   "fell": 0,
+   "replay_s": 43.08,
+   "bad_trajectory": 0,
+   "metric_distance_m": 4.541,
+   "metric_time_s": 71.89,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 99,
+   "agent_resets": 0,
+   "agent_replays": 10,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.273,
+   "peak_processes": 19,
+   "peak_disk_gb": 1.194,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-balance-beam-i00-privileged--fj5xcvd/replay.mp4",
+  "duration_s": 71.92,
+  "fps": 25.0,
+  "frames": 1798,
+  "sha256": "0294c5e6eb85c070ac9ddb18deb44dc6078fbd35393b6110949678246e5a13fd"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "beam_crossed_stable": {
+    "method": "visual",
+    "intervals": [
+     [
+      71.89,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 71.89,
+      "log_step": null,
+      "text_zh": "观察：视频 62-71.8 s 机器人沿梁前移，视频 64 s 起脚在黄色远端标记附近，68-71.8 s 双脚停在黄区、躯干近乎竖直、不再前进，全程未落地、未碰支架。评分器字段：success=1，stages_reached 3/3，fell=0，metric_time_s=71.89（仿真秒，与视频 71.92 s 末尾吻合）。推断：站稳窗口在视频末尾才满，完成时刻取 ≈71.9 s，精度约 ±2 s（站稳窗口约 2 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人左右脚交替踩在窄梁顶面承重行进，最后双脚站在黄色远端；视频中未见手或躯干触梁。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance_m",
+   "value": 4.541,
+   "unit": "m",
+   "threshold": 4.5,
+   "direction": "higher_better"
+  },
+  {
+   "name": "time_s",
+   "value": 71.89,
+   "unit": "s",
+   "threshold": 90.0,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：G1 机器人在视频 0-71.9 s 沿梁走完，双脚站到黄色远端并站稳（评分器 3/3 阶段，fell=0，distance 4.541 m ≥ 4.5，time 71.89 s）。目标节点完成时刻约在视频末尾 71.9 s，由评分器 metric_time 与末帧推断；无状态日志，未逐帧核对接触点。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "成功运行，目标节点 true、视频约 71.9 s 完成，与评分器 3/3、fell=0 及末帧机器人站在梁远端一致；接触标注合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:55:56Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-balance-beam-i00-privileged--fj5xcvd/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-balance-beam-i00-privileged--fj5xcvd/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "c32a5e066d48fe4106233877d168bc34aa45c2d3b3c07652ba887bce44bd6dbd",
+   "video_sha256": "0294c5e6eb85c070ac9ddb18deb44dc6078fbd35393b6110949678246e5a13fd",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/balance_board/trajectories/embodied-olympics-fun-balance-board-i00-standard--m2obhck.json b/phaseR/embodied-olympics-fun/balance_board/trajectories/embodied-olympics-fun-balance-board-i00-standard--m2obhck.json
new file mode 100644
index 0000000000000000000000000000000000000000..f7dd6797cdc91bc7e518715dd482157afde3eb56
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/balance_board/trajectories/embodied-olympics-fun-balance-board-i00-standard--m2obhck.json
@@ -0,0 +1,204 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-balance-board-i00-standard--m2obhck",
+ "taskKey": "embodied-olympics-fun/balance_board",
+ "dag_sha256": "33a43b602711eb1f0ce8accf150479b69fb8b84456d1d75783d478a23c0f1d05",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "calls": 3,
+   "rows": 0,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7194.4,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 4,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.123,
+   "peak_processes": 13,
+   "peak_disk_gb": 2.28,
+   "peak_cpus": 2.31,
+   "peak_pids": 103,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-balance-board-i00-standard--m2obhck.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "c41779731385c8e4e630db551760d97f15fd4b6d807a07f4b81a6a8a8617ceba"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "balanced_30s": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频仅 1 帧（0.04 s），显示 G1 站在滚筒板上的初始姿态，没有任何 30 秒平衡过程；run.json 的 reward 显示 n_actions 0、rows 0、control_steps 0、sim_steps 0，评分器判 success 0。推断：图中只是起始状态，节点未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "观察：reward 中 n_actions 0、rows 0，视频仅 1 帧初始画面；没有轨迹被提交到现场机器人，无接触可记录。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_on_board_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 30.0,
+   "direction": "higher_better"
+  },
+  {
+   "name": "max_tilt_deg",
+   "value": null,
+   "unit": "deg",
+   "threshold": null,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "balanced_30s",
+   "where": "首步前",
+   "t": 0.0,
+   "log_step": null,
+   "evidence_zh": "观察：run.json reward 中 n_actions 0、rows 0、sim_steps 0，log 中 service status control_steps 恒为 0（如 log step 4、21）；视频只有 1 帧起始画面。agent 在约 7200 s 的预算内一直在自己的离线 MuJoCo 副本里调试控制器（log step 240-278），最后一条 say 仍是在等待离线任务（log step 278），没有向现场机器人发送任何动作。推断：平衡步骤的工作从未在评分回合中开始。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "无交付（n_actions 0、control_steps 0），现象与根因留空（D23 Now (a)）。视频仅 1 帧，无状态日志，无法观察平衡过程。根因留空：没尝试和未交付本身不能推出根因；agent 的离线调试只是会话证据，未发现证明其计划错误或控制失败的评分回合证据。时间预算用尽不作为现象或根因。",
+  "unmet": {
+   "balanced_30s": false
+  }
+ },
+ "summary_zh": "观察：评分回合没有任何动作（n_actions 0，control_steps 0），视频仅 1 帧起始画面；唯一节点“平衡满30秒”未达成（0/1）。断点在平衡步骤的首步前，现象留空（D23）；agent 把约 7200 s 都用在离线仿真调试上，未向现场机器人发送动作，根因证据不足故留空。证据缺口：无状态日志、视频无过程。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点 balanced_30s/首步前、时钟都正确（视频仅1帧0.04 s，control_steps恒为0，agent在7174 s仍在离线调试）。唯一问题：这是『agent 侧原因、无任何交付』的 run，DECISIONS.md D23 的现行规则是 phenomena 与 root_causes 都留空并写 gap_zh，不填没尝试；文件已留空根因，但仍填了没尝试。",
+    "n_issues": 1,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "文件已按上轮意见改为现象/根因留空并写 gap_zh（D23）；节点未达成、断点 balanced_30s/首步前、时钟均与证据一致（视频仅1帧0.04 s，n_actions 0，control_steps 0）。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:50Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-balance-board-i00-standard--m2obhck/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-balance-board-i00-standard--m2obhck/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-balance-board-i00-standard--m2obhck/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-balance-board-i00-standard--m2obhck/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "963ca7ae44f1f456b2e581583ac6bde95eb2fdbcacff54e46955979335829189",
+   "video_sha256": "c41779731385c8e4e630db551760d97f15fd4b6d807a07f4b81a6a8a8617ceba",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/basketball_dribble/trajectories/embodied-olympics-fun-basketball-dribble-i00-standard--we2bodk.json b/phaseR/embodied-olympics-fun/basketball_dribble/trajectories/embodied-olympics-fun-basketball-dribble-i00-standard--we2bodk.json
new file mode 100644
index 0000000000000000000000000000000000000000..cd787ffa97d61ef7877bfd3c8cd6c02afc4d4b73
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/basketball_dribble/trajectories/embodied-olympics-fun-basketball-dribble-i00-standard--we2bodk.json
@@ -0,0 +1,221 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-basketball-dribble-i00-standard--we2bodk",
+ "taskKey": "embodied-olympics-fun/basketball_dribble",
+ "dag_sha256": "d5a4c6c70a883e3bab379df3e40c39ce5d4c704f8f7d56d8b3a445f58ada085b",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 610,
+   "sim_steps": 3253,
+   "calls": 34,
+   "rows": 3253,
+   "resets": 4,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7288.1,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.286,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.372,
+   "peak_cpus": 2.89,
+   "peak_pids": 108,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-basketball-dribble-i00-standard--we2bodk.mp4",
+  "duration_s": 6.12,
+  "fps": 25.0,
+  "frames": 153,
+  "sha256": "ce663b310406b10eb07e45a3e1b78d8acfb76ac753b8455d86b9f23a275aff48"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "dribble_started": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 3.0,
+      "log_step": null,
+      "text_zh": "观察：视频 0-3 s 机器人在罚球区右下角原地站立，球从胸前落下，在脚边弹跳后停在身前地面，没有被单手续拍的画面；约 4.5 s 起机器人下蹲/前倾去够球，球仍在脚边。仅松手掉球不计开始运球。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "观察：视频 0-1 s 双手松开球后，机器人手未见有效单手拍球；约 4.5 s 后机器人弯腰用手臂去推/拍脚边的球，球未被续拍成功（接触不清晰）。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：双脚站立于原地支撑，未行走。"
+     }
+    ]
+   },
+   "free_throw_line_passed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程（至 6.12 s）机器人和球始终在起点罚球区附近，未越过罚球线/中线；前置节点未达成。"
+     }
+    ]
+   },
+   "frontcourt_reached": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 6.0,
+      "log_step": null,
+      "text_zh": "观察：视频全程（至 6.12 s）机器人和球始终在起点罚球区附近，未越过罚球线/中线；前置节点未达成。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": "higher_better"
+  },
+  {
+   "name": "dribbles",
+   "value": null,
+   "unit": "count",
+   "threshold": 1,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "dribble_started",
+   "t": 1.5,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0-3 s 球落地后在脚边小幅弹跳后停住，没有合法的单手续拍；推断：首个运球触球未形成，后续节点从未开始。评分器 first_reason_code=2（具体含义 DAG 未解释）；log step 276 的一次测试曾报告“球碰到了手以外的部位”（agent 会话证据）。",
+   "log_step": null
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没建立接触",
+    "t": 2.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频约 0.5-3 s 球落地弹跳后，机器人的手没有在球回升时拍球，球停在脚边。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "无状态日志，视频视角远且球很小，无法确认球是否与手/身体接触；根因无直接证据（agent 会话中大量调试，未交付可验证的运球策略），故根因留空。first_reason_code=2 的含义未知。",
+  "unmet": {
+   "dribble_started": false,
+   "free_throw_line_passed": false,
+   "frontcourt_reached": false
+  }
+ },
+ "summary_zh": "观察：graded 视频 6.12 s，机器人始终在起点罚球区附近，球落地后停在脚边，三个节点均未达成（0/3）。断点为开始运球：首次离手后没有合法的单手续拍。无状态日志且画面小，根因不明。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "视频 6.12 s 内机器人始终停在起点罚球区，球落地弹跳后停在脚边，未见手的续拍；三个节点均未达成、断点在 dribble_started、现象“没建立接触”与根因留空（附 gap_zh）均与画面一致。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:28Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-basketball-dribble-i00-standard--we2bodk/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-basketball-dribble-i00-standard--we2bodk/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f6cbf0373006be8f2025d555eb402e96b4e42715f445161455a1db012ea8c375",
+   "video_sha256": "ce663b310406b10eb07e45a3e1b78d8acfb76ac753b8455d86b9f23a275aff48",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/box_jump/trajectories/embodied-olympics-fun-box-jump-i00-privileged--o9u84u8.json b/phaseR/embodied-olympics-fun/box_jump/trajectories/embodied-olympics-fun-box-jump-i00-privileged--o9u84u8.json
new file mode 100644
index 0000000000000000000000000000000000000000..a11b319aa92c4e61eecccd2290596e37396423d4
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/box_jump/trajectories/embodied-olympics-fun-box-jump-i00-privileged--o9u84u8.json
@@ -0,0 +1,146 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-box-jump-i00-privileged--o9u84u8",
+ "taskKey": "embodied-olympics-fun/box_jump",
+ "dag_sha256": "94c6447a4b03c43587761e86e851339c4316a0cf5855b294978de845b6362c39",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 365,
+   "sim_steps": 365,
+   "horizon": 2000,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 3,
+   "stages_total": 3,
+   "fell": 0,
+   "replay_s": 4.88,
+   "bad_trajectory": 0,
+   "metric_air_s": 0.514,
+   "metric_pelvis_above_box_m": 1.027,
+   "metric_pelvis_rise_m": 0.531,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 49,
+   "agent_resets": 9864,
+   "agent_replays": 5,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-box-jump-i00-privileged--o9u84u8/replay.mp4",
+  "duration_s": 3.68,
+  "fps": 25.0,
+  "frames": 92,
+  "sha256": "2dd766fbd55a3c624a85636ae1861f1ac7ed267347b6d777569cdfb4eb22eb90"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "box_jump_stood": {
+    "method": "visual",
+    "intervals": [
+     [
+      3.6,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 3.6,
+      "log_step": null,
+      "text_zh": "观察：视频约 0.75 s 机器人双脚离地、腾空越过箱前；约 1.0 s 双脚落在箱顶；1.5-3.68 s 一直站直在箱顶，无其他部位触箱/触地，最后一帧仍站立。评分器字段 success=1、stages_reached 3/3、fell=0、air_s 0.514、pelvis_above_box 1.027 m。推断：站直 200 控制步窗口在视频约 3.6 s（回放 365 步=3.65 s）完成，精确时刻不确定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在地面蹬起，腾空后双脚落到箱顶并站直承重；手臂不触箱。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": null,
+ "summary_zh": "成功：机器人（agent 脚本生成的 365 步动作回放）双脚蹬起腾空越过箱前，落在箱顶并站直保持到视频结束（3.68 s），唯一目标节点达成。评分器 success=1，3/3 阶段。站直完成时刻约 3.6 s 由回放步数推断；无状态日志。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "单节点 box_jump_stood 达成，视频约 0.75 s 腾空、约 1.0 s 落箱顶、之后站直至结束，与评分器 success=1、3/3 阶段一致；3.6 s 由回放步数推断且已标注。无失败层，无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:55:38Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-box-jump-i00-privileged--o9u84u8/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-box-jump-i00-privileged--o9u84u8/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "a87d75b34be6e2792b1f3697e82c913a019a904f3ab055b2a2090afab4795d8f",
+   "video_sha256": "2dd766fbd55a3c624a85636ae1861f1ac7ed267347b6d777569cdfb4eb22eb90",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/catch/trajectories/embodied-olympics-fun-catch-i00-realtime--tsfcjde.json b/phaseR/embodied-olympics-fun/catch/trajectories/embodied-olympics-fun-catch-i00-realtime--tsfcjde.json
new file mode 100644
index 0000000000000000000000000000000000000000..ff8880684dd2115f352efb6e61b7cb829f7b4209
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/catch/trajectories/embodied-olympics-fun-catch-i00-realtime--tsfcjde.json
@@ -0,0 +1,235 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-catch-i00-realtime--tsfcjde",
+ "taskKey": "embodied-olympics-fun/catch",
+ "dag_sha256": "61132849363e640170264bbd47c4c4cebc23f7ecfe683cae11a4abb07d10d44b",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 404,
+   "sim_steps": 404,
+   "calls": 10,
+   "rows": 0,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 509.9,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "realtime": 1,
+   "slowdown": 1.0,
+   "clock_s": 7.592,
+   "max_lag_s": 0.512,
+   "stalls": 7,
+   "stalled_s": 3.542,
+   "stalled_share": 0.4665,
+   "held_steps": 404,
+   "late_steps": 0,
+   "requests_refused": 0,
+   "request_busy_s": 0.457,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.085,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.452,
+   "peak_cpus": 1.41,
+   "peak_pids": 146,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-catch-i00-realtime--tsfcjde.mp4",
+  "duration_s": 4.08,
+  "fps": 25.0,
+  "frames": 102,
+  "sha256": "4f82f07b8d0e1a12dc7b6ffae43bb656bc25f98f385780a313ced2687577c566"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "fly_ball_caught": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 4.0,
+      "log_step": 30,
+      "text_zh": "观察：视频 0-4.08 s 机器人几乎原地站立、双手前伸，没有位移或接球动作；球约视频 3.7 s 才在远处小点出现，约 4.0 s 落在机器人右侧地面草坪上，没有进入任何一只手。评分器 first_success=0、success=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "其他",
+      "note_zh": "观察：视频 0-4.08 s 球没有碰到任何一只手；双手前伸张开，球落到地面，未形成抓握或其他接触（候选：无接触）。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚站立支撑，基本未移动。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "catch_height_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": null
+  },
+  {
+   "name": "catch_speed_m_s",
+   "value": null,
+   "unit": "m/s",
+   "threshold": null,
+   "direction": null
+  },
+  {
+   "name": "landed_from_start_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "fly_ball_caught",
+   "t": 4.0,
+   "log_step": 30,
+   "where": "首步前",
+   "evidence_zh": "观察：评分局只含 log step 30 的 get_obs(seconds=5.0) 被动录像（agent 自称 undisturbed ball flight），服务读数 resets=0、held_steps=404=n_actions，未发任何关节指令；视频 0-4.08 s 机器人姿态不变，球约 3.7-4.0 s 落在其右侧地面，接球步骤未启动。"
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": 30,
+    "evidence_zh": "观察：评分局只含 log step 30 的 get_obs(seconds=5.0) 被动录像，服务读数 resets=0、held_steps=404=n_actions，未发任何关节指令；视频 0-4.08 s 机器人姿态不变，球约 3.7-4.0 s 落地。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "根因留空：评分局只是 agent 的 get_obs(5 s) 被动录像（log step 30，resets=0），未发指令；其后的 mujoco 离线仿真（log 42-296）从未送入服务，不是评分局证据。无状态日志；native_metrics 在 reward 中未报告；视频中球小且远，无法判断球的真实落点。",
+  "unmet": {
+   "fly_ball_caught": false
+  }
+ },
+ "summary_zh": "观察：graded 视频 4.08 s 中机器人站立不动，球在约视频 3.7 s 才出现并落在其右侧地面，接球节点未达成（0/1）。断点为接稳飞球·首步前，现象为没尝试（评分局只是被动录像，未发指令）；根因未能判定。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 8,
+    "summary_zh": "评分局里机器人从未收到任何控制指令：agent 用唯一一次 get_obs(seconds=5) 把整局当成“无干扰录像”跑完（run.json log step 30，resets=0，held_steps=404=n_actions），所以机器人原地不动是“没尝试”，不是“动态时机”。断点 ref 正确，但 where 应为首步前，现象应改为 没尝试。",
+    "n_issues": 3,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "第二轮：修改后的标签与证据一致。评分局只是 get_obs(5 s) 被动录像，机器人未动，球落在右侧草坪；节点未达成，断点首步前、没尝试，根因留空有 gap_zh，符合 FAIL-22。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:13Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-catch-i00-realtime--tsfcjde/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-catch-i00-realtime--tsfcjde/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-catch-i00-realtime--tsfcjde/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-catch-i00-realtime--tsfcjde/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "166cfa71b1cb69f471810c29e1a491e514af46a16efb52d87342dd9d081ef57b",
+   "video_sha256": "4f82f07b8d0e1a12dc7b6ffae43bb656bc25f98f385780a313ced2687577c566",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/catch_and_shoot/trajectories/embodied-olympics-fun-catch-and-shoot-i00-standard--nt5twet.json b/phaseR/embodied-olympics-fun/catch_and_shoot/trajectories/embodied-olympics-fun-catch-and-shoot-i00-standard--nt5twet.json
new file mode 100644
index 0000000000000000000000000000000000000000..05ccbb9c450932ae1c11dc69bded95646dbdafb3
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/catch_and_shoot/trajectories/embodied-olympics-fun-catch-and-shoot-i00-standard--nt5twet.json
@@ -0,0 +1,187 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-catch-and-shoot-i00-standard--nt5twet",
+ "taskKey": "embodied-olympics-fun/catch_and_shoot",
+ "dag_sha256": "f6a6d84d578d7c1084b9432c694b8d43d8fa3421ae8d0142dbb9a316794fb326",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 267,
+   "sim_steps": 267,
+   "calls": 13,
+   "rows": 267,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 5798.5,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.291,
+   "peak_processes": 11,
+   "peak_disk_gb": 2.424,
+   "peak_cpus": 2.31,
+   "peak_pids": 137,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-catch-and-shoot-i00-standard--nt5twet.mp4",
+  "duration_s": 2.68,
+  "fps": 25.0,
+  "frames": 67,
+  "sha256": "687d440daed01ef8e1e0817b16fd0b029d9755254dc51ba4a3753140c36b7403"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": null,
+  "peak": null,
+  "final": null,
+  "nodes": {
+   "ball_caught": {
+    "method": "unknown",
+    "intervals": null,
+    "final": null,
+    "evidence": [
+     {
+      "t": 2.45,
+      "log_step": null,
+      "text_zh": "观察：视频 2.0-2.3 s 球从左侧传球机飞向机器人；视频 2.4-2.68 s 球出现在机器人胸前双手附近，被运动模糊和手指遮挡，看不清两手是否同时持续接触 0.10 s；视频在 2.68 s 结束，无投篮画面。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "球在视频约 2.4 s 起接近双手，遮挡严重，无法确认两手接触方式。"
+     }
+    ]
+   },
+   "shot_through_ring": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.68,
+      "log_step": null,
+      "text_zh": "观察：视频至 2.68 s 结束，球始终未离开机器人双手，没有出手和飞向篮圈的画面；评分 success=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "未出手，未进行投篮步骤。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "ball_caught",
+   "where": "操作中",
+   "t": 2.4,
+   "evidence_zh": "观察：视频 2.4-2.68 s 球到达双手处后视频结束，没有出手；评分 success=0（first_reason_code 2，具体含义未提供）。推断：失败发生在接球步骤，未能确认其完成。",
+   "log_step": null
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "视频仅 2.68 s，球进入双手后被遮挡，无法判断是没接住、接了没保持住还是接住后未能出手；无状态日志；评分器未给出失败原因文字（仅 first_reason_code 2）。agent 日志是整个会话的本地仿真调试，不作为图证。因此现象和根因留空。",
+  "unmet": {
+   "ball_caught": null,
+   "shot_through_ring": false
+  }
+ },
+ "summary_zh": "观察：graded 视频仅 2.68 s，球在约 2.4 s 到达机器人双手处，之后视频结束，未见投篮，投篮命中未达成。接球节点因运动模糊遮挡无法确认，标 unknown；断点定在接球步骤。现象与根因因证据不足留空。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "视频仅 2.68 s，球约 2.4 s 到达双手处后视频结束，无出手；接球 unknown、投篮 false、断点 ball_caught、现象/根因留空并附 gap_zh，均与可见证据和可用来源一致，未发现需要修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:23Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-catch-and-shoot-i00-standard--nt5twet/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-catch-and-shoot-i00-standard--nt5twet/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "de681cdf572becc49ec9e31f4ac5115445056a8d57b91ecaf436a033ca1725e9",
+   "video_sha256": "687d440daed01ef8e1e0817b16fd0b029d9755254dc51ba4a3753140c36b7403",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/clean_and_jerk/trajectories/embodied-olympics-fun-clean-and-jerk-i00-standard--cgna7dx.json b/phaseR/embodied-olympics-fun/clean_and_jerk/trajectories/embodied-olympics-fun-clean-and-jerk-i00-standard--cgna7dx.json
new file mode 100644
index 0000000000000000000000000000000000000000..372f4f7a0bd7246f33dcdea36fbada219379264f
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/clean_and_jerk/trajectories/embodied-olympics-fun-clean-and-jerk-i00-standard--cgna7dx.json
@@ -0,0 +1,228 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-clean-and-jerk-i00-standard--cgna7dx",
+ "taskKey": "embodied-olympics-fun/clean_and_jerk",
+ "dag_sha256": "f530c1d45bf82223dabd69c9d30cf24e240db1d588782640954595693a7015ca",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 1587,
+   "sim_steps": 1894,
+   "calls": 8,
+   "rows": 1557,
+   "resets": 1,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 4297.9,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.211,
+   "peak_processes": 7,
+   "peak_disk_gb": 0.346,
+   "peak_cpus": 4.02,
+   "peak_pids": 143,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-clean-and-jerk-i00-standard--cgna7dx.mp4",
+  "duration_s": 15.88,
+  "fps": 25.0,
+  "frames": 397,
+  "sha256": "5d45333984048f9e9335ad2e15e2c656be05ae054912192ef333463730f00504"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 2,
+  "final": 2,
+  "nodes": {
+   "bar_at_shoulders": {
+    "method": "visual",
+    "intervals": [
+     [
+      5.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 5.0,
+      "log_step": null,
+      "text_zh": "观察：视频约 3.5 s 机器人双手把杠铃从架上提起，4.5-6.5 s 杠铃停在胸肩前方、机器人直立，此后上举不撤销；时间约 ±0.5 s。评分器 success=1，4/4 阶段（log step 177 服务状态 reason=success）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "双臂协作",
+      "how": "抓握",
+      "note_zh": "观察：左右灵巧手各握杠铃一侧，双手同时拉起杠铃至肩前。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：双脚站在举重台上承重，保持直立。"
+     }
+    ]
+   },
+   "bar_overhead": {
+    "method": "visual",
+    "intervals": [
+     [
+      8.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 8.5,
+      "log_step": null,
+      "text_zh": "观察：视频约 7.5 s 起双臂上推，8.5 s 起杠铃在头顶上方、双臂伸直，直到视频结束 15.88 s 保持；时间约 ±0.5 s。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "双臂协作",
+      "how": "抓握",
+      "note_zh": "观察：双手握杠上推并在头顶托住杠铃。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：双脚在木台上并立支撑。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "bar_height_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": "higher_better"
+  },
+  {
+   "name": "bar_max_height_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": "higher_better"
+  },
+  {
+   "name": "elbow_flex_deg",
+   "value": null,
+   "unit": "deg",
+   "threshold": 10.0,
+   "direction": "lower_better"
+  },
+  {
+   "name": "knee_flex_deg",
+   "value": null,
+   "unit": "deg",
+   "threshold": 10.0,
+   "direction": "lower_better"
+  },
+  {
+   "name": "time_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 60.0,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：杠铃约视频 5.0 s 上肩、约 8.5 s 过顶并保持到 15.88 s 结束，2/2 节点完成，评分器 success=1。节点时间由视频目测（无状态日志）；评分器未给出原生指标数值。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点时间与接触与视频一致：约5 s上肩、约8.5 s过顶并保持到结束，评分器success=1，无失败层。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:55:57Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-clean-and-jerk-i00-standard--cgna7dx/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-clean-and-jerk-i00-standard--cgna7dx/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "da08276fe239249f8147ab0097f0585d93289355596086484a932cbbf3c1df43",
+   "video_sha256": "5d45333984048f9e9335ad2e15e2c656be05ae054912192ef333463730f00504",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/discus/trajectories/embodied-olympics-fun-discus-i00-standard--edk3gpq.json b/phaseR/embodied-olympics-fun/discus/trajectories/embodied-olympics-fun-discus-i00-standard--edk3gpq.json
new file mode 100644
index 0000000000000000000000000000000000000000..94d6336c3ffc8c6d9cac45b15ac0098bcc78caa0
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/discus/trajectories/embodied-olympics-fun-discus-i00-standard--edk3gpq.json
@@ -0,0 +1,203 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-discus-i00-standard--edk3gpq",
+ "taskKey": "embodied-olympics-fun/discus",
+ "dag_sha256": "ffb63c003bdbc581b50872ffe890e6ce579309876e24b407597d79050cb48e95",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "calls": 1,
+   "rows": 0,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7181.7,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 4,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.532,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.353,
+   "peak_cpus": 1.31,
+   "peak_pids": 144,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-discus-i00-standard--edk3gpq.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "5d9c1503eb2984af754c58117202641d12a476911fe2c89c8ad2ffe0fad0599d"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "discus_landed_qualified": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：replay.mp4 只有 1 帧（0.04 s），机器人站在投掷圈内，铁饼在右手中，尚未投掷。读取：reward n_actions=0、rows=0、sim_steps=0，agent 没有向机器人服务发送任何动作；推断：铁饼落区达标 未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "没有交付任何轨迹（n_actions 0），机器人未动。"
+     }
+    ]
+   },
+   "rear_exit_stood": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：replay.mp4 只有 1 帧（0.04 s），机器人站在投掷圈内，铁饼在右手中，尚未投掷。读取：reward n_actions=0、rows=0、sim_steps=0，agent 没有向机器人服务发送任何动作；推断：后圈离圈站稳 未达成。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 15.0,
+   "direction": "higher_better"
+  },
+  {
+   "name": "retouch_dv_m_s",
+   "value": null,
+   "unit": "m/s",
+   "threshold": 0.5,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "discus_landed_qualified",
+   "t": 0.0,
+   "log_step": null,
+   "where": "首步前",
+   "evidence_zh": "观察：视频仅 1 帧，机器人站在圈内未动；评分字段 n_actions=0、rows=0，未交付任何动作。log 显示 agent 约 7200 s 都在本地 MuJoCo 里做抓握和投掷仿真（如 log step 235-240），最后一步仍是本地仿真诊断，没有向实机服务发送动作。"
+  },
+  "phenomena": [
+   {
+    "cat": "没尝试",
+    "item": "没尝试",
+    "t": 0.0,
+    "log_step": null,
+    "evidence_zh": "观察：视频仅 1 帧（0.04 s），机器人未动、铁饼仍在手中；reward n_actions=0。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "根因留空：没有交付动作是 agent 一侧的原因，agent 把整个预算花在本地仿真调试上，但日志没有直接证据表明其计划有误或控制有某类问题，按规则不从没尝试推根因。视频只有 1 帧，没有投掷过程可看。",
+  "unmet": {
+   "discus_landed_qualified": false,
+   "rear_exit_stood": false
+  }
+ },
+ "summary_zh": "观察：本次运行没有向机器人发送任何动作（n_actions 0，视频仅 1 帧），两个节点（铁饼落区达标、后圈离圈站稳）都未达成。推断：断点是首步前，agent 约 7200 s 的预算都用在本地 MuJoCo 抓握/投掷仿真上，没有交付。证据缺口：无投掷画面，根因无直接证据。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "视频仅 1 帧、n_actions=0，两节点未达成、断点首步前+没尝试、根因留空并写 gap_zh，均符合 03 规则（agent 侧未交付，时限 7200 s 用尽）。未发现需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:08Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-discus-i00-standard--edk3gpq/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-discus-i00-standard--edk3gpq/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "536894c583946c6ca7967a46196f9e7bdbdc59c53c3c5c1089086f3418ba8c30",
+   "video_sha256": "5d9c1503eb2984af754c58117202641d12a476911fe2c89c8ad2ffe0fad0599d",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/dodgeball/trajectories/embodied-olympics-fun-dodgeball-i00-realtime--ztowuv8.json b/phaseR/embodied-olympics-fun/dodgeball/trajectories/embodied-olympics-fun-dodgeball-i00-realtime--ztowuv8.json
new file mode 100644
index 0000000000000000000000000000000000000000..1b3496b5da9b8ff27e0093ad3c52906975a98876
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/dodgeball/trajectories/embodied-olympics-fun-dodgeball-i00-realtime--ztowuv8.json
@@ -0,0 +1,231 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-dodgeball-i00-realtime--ztowuv8",
+ "taskKey": "embodied-olympics-fun/dodgeball",
+ "dag_sha256": "a13a84ada7b474ef0b05b1dd7e90325e49ac88a53ef3656b957cb843db7034f2",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "calls": 1,
+   "rows": 0,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 916.7,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 4,
+   "realtime": 1,
+   "slowdown": 1.0,
+   "clock_s": 0.0,
+   "max_lag_s": 0.0,
+   "stalls": 0,
+   "stalled_s": 0.0,
+   "stalled_share": 0.0,
+   "held_steps": 0,
+   "late_steps": 0,
+   "requests_refused": 0,
+   "request_busy_s": 0.296,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.134,
+   "peak_processes": 8,
+   "peak_disk_gb": 0.247,
+   "peak_cpus": 0.79,
+   "peak_pids": 17,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-dodgeball-i00-realtime--ztowuv8.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "b6abf95b2cf24115ee0745a9246fe3af727ba08c680b1586c51ed77f965685ab"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 5,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "throw_1_dodged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频只有 1 帧（0.04 s），机器人站在场中央，5 个发球机在远端，场上无球飞行；run.json 显示 n_actions=0、sim_steps=0、calls=1（只读了状态），没有提交任何动作，视频中无来球可被躲过，故本球未躲过。"
+     }
+    ]
+   },
+   "throw_2_dodged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频只有 1 帧（0.04 s），机器人站在场中央，5 个发球机在远端，场上无球飞行；run.json 显示 n_actions=0、sim_steps=0、calls=1（只读了状态），没有提交任何动作，视频中无来球可被躲过，故本球未躲过。"
+     }
+    ]
+   },
+   "throw_3_dodged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频只有 1 帧（0.04 s），机器人站在场中央，5 个发球机在远端，场上无球飞行；run.json 显示 n_actions=0、sim_steps=0、calls=1（只读了状态），没有提交任何动作，视频中无来球可被躲过，故本球未躲过。"
+     }
+    ]
+   },
+   "throw_4_dodged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频只有 1 帧（0.04 s），机器人站在场中央，5 个发球机在远端，场上无球飞行；run.json 显示 n_actions=0、sim_steps=0、calls=1（只读了状态），没有提交任何动作，视频中无来球可被躲过，故本球未躲过。"
+     }
+    ]
+   },
+   "throw_5_dodged": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：视频只有 1 帧（0.04 s），机器人站在场中央，5 个发球机在远端，场上无球飞行；run.json 显示 n_actions=0、sim_steps=0、calls=1（只读了状态），没有提交任何动作，视频中无来球可被躲过，故本球未躲过。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "dodged",
+   "value": null,
+   "unit": "count",
+   "threshold": 5,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "unknown",
+   "t": null,
+   "log_step": 10,
+   "where": "首步前",
+   "evidence_zh": "观察：agent 在 log step 10（agent 工作 206 s）以及之后的 13、15、17、19 步均收到 Gemini 模型提供方的内容过滤拦截（“This request was blocked by Gemini's filters”），且 reward 中 n_actions=0、sim_steps=0；视频仅 1 帧，机器人始终站在初始位置。推断：agent 在提交任何动作前被提供方过滤中断，属外部因素，无已交付轨迹。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "外部因素：模型提供方过滤拦截（log step 10、13、15、17、19）在任何轨迹交付前切断了运行，按规则不标现象与根因；视频仅 1 帧（0.04 s），无法观察任何来球。",
+  "unmet": {
+   "throw_1_dodged": false,
+   "throw_2_dodged": false,
+   "throw_3_dodged": false,
+   "throw_4_dodged": false,
+   "throw_5_dodged": false
+  }
+ },
+ "summary_zh": "观察：视频仅 1 帧（0.04 s），机器人未动，5 个躲球节点均未达成（n_actions=0）。agent 在 log step 10 起多次被 Gemini 过滤拦截，未提交任何动作，属外部因素切断，断点 unknown·首步前，不标现象与根因。证据缺口：没有可观察的来球过程。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标签与证据一致：视频仅1帧，n_actions=0、sim_steps=0，5个躲球节点均未达成；agent log step 10/13/15/17/19 均为Gemini过滤拦截，属外部因素，现象与根因为空并有gap_zh，符合D23/D37。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:17Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-dodgeball-i00-realtime--ztowuv8/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-dodgeball-i00-realtime--ztowuv8/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "3a33351bb15f4b41ac8d58f1bc1acc1ae9bae04594e6ef10b6cd7c13c9427fc4",
+   "video_sha256": "b6abf95b2cf24115ee0745a9246fe3af727ba08c680b1586c51ed77f965685ab",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/dodgeball/trajectories/embodied-olympics-fun-dodgeball-i00-standard--j5v3x37.json b/phaseR/embodied-olympics-fun/dodgeball/trajectories/embodied-olympics-fun-dodgeball-i00-standard--j5v3x37.json
new file mode 100644
index 0000000000000000000000000000000000000000..849296dce6ec2849044ee74d993d8343b01df361
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/dodgeball/trajectories/embodied-olympics-fun-dodgeball-i00-standard--j5v3x37.json
@@ -0,0 +1,254 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-dodgeball-i00-standard--j5v3x37",
+ "taskKey": "embodied-olympics-fun/dodgeball",
+ "dag_sha256": "a13a84ada7b474ef0b05b1dd7e90325e49ac88a53ef3656b957cb843db7034f2",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 1334,
+   "sim_steps": 1334,
+   "calls": 6,
+   "rows": 1334,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 624.2,
+   "video_rendered": 1,
+   "first_success": 1,
+   "first_reason_code": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.343,
+   "peak_processes": 6,
+   "peak_disk_gb": 0.263,
+   "peak_cpus": 0.88,
+   "peak_pids": 75,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-dodgeball-i00-standard--j5v3x37.mp4",
+  "duration_s": 13.36,
+  "fps": 25.0,
+  "frames": 334,
+  "sha256": "47f04b3ce62f5ed54014b1855b557521ace62f81b25bc56710f6dcbd9bdf295d"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 5,
+  "base": 0,
+  "peak": 5,
+  "final": 5,
+  "nodes": {
+   "throw_1_dodged": {
+    "method": "visual",
+    "intervals": [
+     [
+      3.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 3.5,
+      "log_step": 31,
+      "text_zh": "观察：场地远景视频中第1个红球飞向机器人后在其身旁落地/成为死球，机器人未被击中、未出界未跌倒（时间约 视频 3.5 s，由 1.5 帧/秒的缩略图估计，精度约 ±1 s）；评分器 reward 记录 success=1，日志 step 31/32 服务读数 Stages 5/5，done reason success。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "避开",
+      "note_zh": "机器人通过移步/转身避开球，未与活球接触"
+     }
+    ]
+   },
+   "throw_2_dodged": {
+    "method": "visual",
+    "intervals": [
+     [
+      6.2,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 6.2,
+      "log_step": 31,
+      "text_zh": "观察：场地远景视频中第2个红球飞向机器人后在其身旁落地/成为死球，机器人未被击中、未出界未跌倒（时间约 视频 6.2 s，由 1.5 帧/秒的缩略图估计，精度约 ±1 s）；评分器 reward 记录 success=1，日志 step 31/32 服务读数 Stages 5/5，done reason success。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "避开",
+      "note_zh": "机器人通过移步/转身避开球，未与活球接触"
+     }
+    ]
+   },
+   "throw_3_dodged": {
+    "method": "visual",
+    "intervals": [
+     [
+      8.2,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 8.2,
+      "log_step": 31,
+      "text_zh": "观察：场地远景视频中第3个红球飞向机器人后在其身旁落地/成为死球，机器人未被击中、未出界未跌倒（时间约 视频 8.2 s，由 1.5 帧/秒的缩略图估计，精度约 ±1 s）；评分器 reward 记录 success=1，日志 step 31/32 服务读数 Stages 5/5，done reason success。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "避开",
+      "note_zh": "机器人通过移步/转身避开球，未与活球接触"
+     }
+    ]
+   },
+   "throw_4_dodged": {
+    "method": "visual",
+    "intervals": [
+     [
+      10.8,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 10.8,
+      "log_step": 31,
+      "text_zh": "观察：场地远景视频中第4个红球飞向机器人后在其身旁落地/成为死球，机器人未被击中、未出界未跌倒（时间约 视频 10.8 s，由 1.5 帧/秒的缩略图估计，精度约 ±1 s）；评分器 reward 记录 success=1，日志 step 31/32 服务读数 Stages 5/5，done reason success。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "避开",
+      "note_zh": "机器人通过移步/转身避开球，未与活球接触"
+     }
+    ]
+   },
+   "throw_5_dodged": {
+    "method": "visual",
+    "intervals": [
+     [
+      13.3,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 13.3,
+      "log_step": 31,
+      "text_zh": "观察：场地远景视频中第5个红球飞向机器人后在其身旁落地/成为死球，机器人未被击中、未出界未跌倒（时间约 视频 13.3 s，由 1.5 帧/秒的缩略图估计，精度约 ±1 s）；评分器 reward 记录 success=1，日志 step 31/32 服务读数 Stages 5/5，done reason success。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "避开",
+      "note_zh": "机器人通过移步/转身避开球，未与活球接触"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": null,
+ "summary_zh": "成功：视频 13.36 s 内五次来球均被躲过（5/5，评分器 success=1，log step 31 服务读数 Stages 5/5）。各球时间由低密度缩略图估计（约±1 s），未逐帧核对；无状态日志。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "成功运行，5 个球节点均 final=true 且时间（视频 3.5/6.2/8.2/10.8/13.3 s）与帧一致（误差≤1 s）；contact_actual 为 腿/躯干·避开 合理；评分器 success=1 支持 5/5，无失败层。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:11Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-dodgeball-i00-standard--j5v3x37/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-dodgeball-i00-standard--j5v3x37/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "90ef0484b8536e3598756ea14b9820359293afb2664bd37713160353712e850d",
+   "video_sha256": "47f04b3ce62f5ed54014b1855b557521ace62f81b25bc56710f6dcbd9bdf295d",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/free_kick_wall/trajectories/embodied-olympics-fun-free-kick-wall-i00-privileged--rrelrzw.json b/phaseR/embodied-olympics-fun/free_kick_wall/trajectories/embodied-olympics-fun-free-kick-wall-i00-privileged--rrelrzw.json
new file mode 100644
index 0000000000000000000000000000000000000000..e8501b315914f04e3afea6e53620196d9b7bb205
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/free_kick_wall/trajectories/embodied-olympics-fun-free-kick-wall-i00-privileged--rrelrzw.json
@@ -0,0 +1,202 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-free-kick-wall-i00-privileged--rrelrzw",
+ "taskKey": "embodied-olympics-fun/free_kick_wall",
+ "dag_sha256": "6749223ae6b8a425b1f7d86513bee960c64b42ed756a390e3a358baebdc119ab",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 359,
+   "sim_steps": 359,
+   "horizon": 3000,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 2,
+   "stages_total": 2,
+   "fell": 0,
+   "replay_s": 4.12,
+   "bad_trajectory": 0,
+   "metric_ball_x_m": 20.498,
+   "metric_cross_y_m": 1.343,
+   "metric_cross_z_m": 0.207,
+   "metric_kick_speed_m_s": 10.48,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 114,
+   "agent_resets": 27512,
+   "agent_replays": 78,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-free-kick-wall-i00-privileged--rrelrzw/replay.mp4",
+  "duration_s": 3.6,
+  "fps": 25.0,
+  "frames": 90,
+  "sha256": "10dbc83d073b55a1f98825e6890fee773cc39bef43a9d45eba110ba9c9266b8f"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "free_kick_scored": {
+    "method": "visual",
+    "intervals": [
+     [
+      0.9,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.9,
+      "log_step": null,
+      "text_zh": "观察：视频约 0.9 s 机器人右脚击球，球向前飞出并落在远离球门处的场地中部偏左，约 1.8 s 起静止，视频中未见球越线（球很小，视频可能与评分轨迹不完全对应）。推断：越线由评分器字段认定——success 1，stages_reached 2/2，metric_cross_y_m 1.343（在 ±3.66 内），metric_cross_z_m 0.207（在 0-2.44 内），ball_x 20.498，kick_speed 10.48 m/s，fell 0；到达时间取击球时刻 0.9 s 附近，仅为估计，非视频观察。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：视频约 0.9 s 机器人脚部击打足球（候选名称：击打）；站立的另一只脚支撑。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：踢球时支撑脚站在草地上。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "ball_x_m",
+   "value": 20.498,
+   "unit": "m",
+   "threshold": 20.49,
+   "direction": "higher_better"
+  },
+  {
+   "name": "cross_y_m",
+   "value": 1.343,
+   "unit": "m",
+   "threshold": [
+    -3.66,
+    3.66
+   ],
+   "direction": "target"
+  },
+  {
+   "name": "cross_z_m",
+   "value": 0.207,
+   "unit": "m",
+   "threshold": [
+    0,
+    2.44
+   ],
+   "direction": "target"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：评分器判定成功（cross_y 1.343 m、cross_z 0.207 m，stages 2/2，未倒地）；视频中只见约 0.9 s 踢球，未见越线，节点时间 0.9 s 为基于评分器的估计。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "final=true 由评分器字段（success 1、stages 2/2、cross_y/z 在范围内）支持，无异议；但证据里“球越过人墙侧方、接近球门并落入球门区域”的视频观察不成立：帧里球踢出后只飞了一小段就落在场地中部偏左，约1.8 s 起停在远离球门和人墙的位置，视频中看不到球越线，1.7 s 的越线时刻没有视频依据。",
+    "n_issues": 3,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "成功由评分器字段（success 1、stages 2/2、cross_y/z 在范围内）支持；第1轮后证据已区分视频观察（约0.9 s 踢球）与评分器推断，时间标注为估计，无新问题。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:01Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-free-kick-wall-i00-privileged--rrelrzw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-free-kick-wall-i00-privileged--rrelrzw/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-free-kick-wall-i00-privileged--rrelrzw/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-free-kick-wall-i00-privileged--rrelrzw/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "588a8dbcf79eb614bd5aaf9f40f541fa00efe2bea3259dc33a2c4f8b2f70022b",
+   "video_sha256": "10dbc83d073b55a1f98825e6890fee773cc39bef43a9d45eba110ba9c9266b8f",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/free_throw/trajectories/embodied-olympics-fun-free-throw-i00-standard--97uncb7.json b/phaseR/embodied-olympics-fun/free_throw/trajectories/embodied-olympics-fun-free-throw-i00-standard--97uncb7.json
new file mode 100644
index 0000000000000000000000000000000000000000..28b1b937dbd795c1d68ace0251f079d54c1ad2ba
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/free_throw/trajectories/embodied-olympics-fun-free-throw-i00-standard--97uncb7.json
@@ -0,0 +1,210 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-free-throw-i00-standard--97uncb7",
+ "taskKey": "embodied-olympics-fun/free_throw",
+ "dag_sha256": "c2d4b07a9ca0ff585b81cafb3eecf9f2484164969c066ba3cd5ef111abe016a5",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 132,
+   "sim_steps": 261,
+   "calls": 4,
+   "rows": 261,
+   "resets": 1,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 6847.6,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.087,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.417,
+   "peak_cpus": 1.94,
+   "peak_pids": 144,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-free-throw-i00-standard--97uncb7.mp4",
+  "duration_s": 1.36,
+  "fps": 25.0,
+  "frames": 34,
+  "sha256": "e57284622e4b5a3be894d3e344057a7dd98ec4e610b170a4bf90a2460d3c0808"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "free_throw_scored": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.3,
+      "log_step": null,
+      "text_zh": "观察：视频 0.2 s 球离手后沿低弧线朝篮架方向飞行，约 0.9 s 时在篮板下方、低于圈高，1.2 s 落在篮架立柱底部附近，未升到圈高，无穿圈。评分器 success=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "双臂协作",
+      "how": "托/舀",
+      "note_zh": "观察：视频 0-0.2 s 双手托球上举后送出；腿/躯干站立支撑未单列"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：双脚站在罚球线后，全程未倒"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "ball_height_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "free_throw_scored",
+   "t": 0.3,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.2 s 球离手后沿低弧线朝篮架方向飞行，约 0.9 s 时已在篮板下方、低于圈高，1.2 s 落在篮架立柱底部附近，未到圈高，未穿圈。"
+  },
+  "unmet": {
+   "free_throw_scored": false
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.3,
+    "log_step": null,
+    "evidence_zh": "观察：视频 0.3-1.2 s 球向篮架飞行但弹道偏低，未升到篮圈高度即下落，落在篮架下方地面；推断为出手速度/仰角不足。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "无状态日志，球的出手速度/角度只能从 25 fps 的 34 帧估计；agent 最后在 log step 279-282 仍在调试出手参数（自述，非视频证据），不足以指定根因子类，故根因留空。"
+ },
+ "summary_zh": "观察：视频仅 1.36 s，机器人双手托球后于约 0.2 s 抛出，球沿低弧线飞向篮架，但始终低于圈高，约 1.2 s 落在立柱底部附近，罚球命中节点未达成。断在出手本身（推断为速度/仰角不足），无状态日志，根因证据不足，留空。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点 free_throw_scored 与 root_causes 留空均可接受；但证据把球轨迹写成「近乎竖直、落在罚球线附近」与视频不符：球实际朝篮筐方向飞出约一个球场纵深，但始终低于篮圈高度，在篮架立柱底部附近落地（出手速度/仰角不足，未到圈高）。需改写 breakpoint/phenomena/summary/节点证据中的轨迹描述。",
+    "n_issues": 2,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点 free_throw_scored、位姿不准现象与空根因（带 gap_zh）均与视频相符；球低弧线飞向篮架、低于圈高、落在立柱底部附近。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:58:56Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-free-throw-i00-standard--97uncb7/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-free-throw-i00-standard--97uncb7/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-free-throw-i00-standard--97uncb7/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-free-throw-i00-standard--97uncb7/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "be8d5d62ac6673e0fc87eb0da5b748808644b8273261049597aa449d7ab216a1",
+   "video_sha256": "e57284622e4b5a3be894d3e344057a7dd98ec4e610b170a4bf90a2460d3c0808",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/goalkeeper_save/trajectories/embodied-olympics-fun-goalkeeper-save-i00-standard--zmzuanh.json b/phaseR/embodied-olympics-fun/goalkeeper_save/trajectories/embodied-olympics-fun-goalkeeper-save-i00-standard--zmzuanh.json
new file mode 100644
index 0000000000000000000000000000000000000000..5c3e058e5cdc8342096facf3ede293c59a9468fa
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/goalkeeper_save/trajectories/embodied-olympics-fun-goalkeeper-save-i00-standard--zmzuanh.json
@@ -0,0 +1,215 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-goalkeeper-save-i00-standard--zmzuanh",
+ "taskKey": "embodied-olympics-fun/goalkeeper_save",
+ "dag_sha256": "8e220a4494718f032caa453eced9c614dcf4d2132dc0c7d4d5a8b7a9cfdc58d1",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 2670,
+   "sim_steps": 6336,
+   "calls": 455,
+   "rows": 6336,
+   "resets": 4,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 5943.4,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.566,
+   "peak_processes": 10,
+   "peak_disk_gb": 1.309,
+   "peak_cpus": 6.49,
+   "peak_pids": 136,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_cpus": 1
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-goalkeeper-save-i00-standard--zmzuanh.mp4",
+  "duration_s": 26.72,
+  "fps": 25.0,
+  "frames": 668,
+  "sha256": "cd2bbf450b2b76d2f47ee9b604834ef0817c56b8cf353a8b29b8cc32a48d0550"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 3,
+  "final": 3,
+  "nodes": {
+   "save_1_completed": {
+    "method": "visual",
+    "intervals": [
+     [
+      8.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 8.5,
+      "log_step": null,
+      "text_zh": "观察：视频约 6.0 s 起可见发球机的球飞向球门（2 fps 抽帧），机器人守门员站在球门内；之后球门内无进球。推断：开球后满 2.5 s 无进球，故裁判锁存于视频约 8.5 s（时间为推算，约±1 s）。评分器字段 success=1，reward 记录 3/3 阶段（agent 末尾自述 STAGES: 3 / 3，log 末步）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚站在球门线附近草地上。球是否碰到机身或手臂在 640×360 画面中无法分辨，故不确认击球接触。"
+     }
+    ]
+   },
+   "save_2_completed": {
+    "method": "visual",
+    "intervals": [
+     [
+      17.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 17.5,
+      "log_step": null,
+      "text_zh": "观察：视频约 15.0 s 起可见发球机的球飞向球门（2 fps 抽帧），机器人守门员站在球门内；之后球门内无进球。推断：开球后满 2.5 s 无进球，故裁判锁存于视频约 17.5 s（时间为推算，约±1 s）。评分器字段 success=1，reward 记录 3/3 阶段（agent 末尾自述 STAGES: 3 / 3，log 末步）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚站在球门线附近草地上。球是否碰到机身或手臂在 640×360 画面中无法分辨，故不确认击球接触。"
+     }
+    ]
+   },
+   "save_3_completed": {
+    "method": "visual",
+    "intervals": [
+     [
+      26.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 26.5,
+      "log_step": null,
+      "text_zh": "观察：视频约 24.0 s 起可见发球机的球飞向球门（2 fps 抽帧），机器人守门员站在球门内；之后球门内无进球。推断：开球后满 2.5 s 无进球，故裁判锁存于视频约 26.5 s（时间为推算，约±1 s）。评分器字段 success=1，reward 记录 3/3 阶段（agent 末尾自述 STAGES: 3 / 3，log 末步）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚站在球门线附近草地上。球是否碰到机身或手臂在 640×360 画面中无法分辨，故不确认击球接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "saved",
+   "value": 3,
+   "unit": "次",
+   "threshold": 3,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：三次扑救节点均达成（视频约 8.5、17.5、26.5 s，由球出现时刻 +2.5 s 推算，第三次接近视频末 26.72 s）。评分器 success=1，三阶段全过。缺口：全景分辨率低，看不清球与机器人的具体接触，也无状态日志；各节点时间为推算值。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "成功运行，三个事件节点均达成，final=3/3 与评分器 success=1 一致；发球时刻（视频约 6、15、24 s）与抽帧所见一致，+2.5 s 推算已标为推断，接触未知处如实说明。未发现需改动的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:35Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-goalkeeper-save-i00-standard--zmzuanh/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-goalkeeper-save-i00-standard--zmzuanh/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "c11299529f3ecf3b8791f2f530ee24cedfaefbacbc8f0e0eb640b75f52373042",
+   "video_sha256": "cd2bbf450b2b76d2f47ee9b604834ef0817c56b8cf353a8b29b8cc32a48d0550",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/hurdles_freestanding/trajectories/embodied-olympics-fun-hurdles-freestanding-i00-standard--wzyztys.json b/phaseR/embodied-olympics-fun/hurdles_freestanding/trajectories/embodied-olympics-fun-hurdles-freestanding-i00-standard--wzyztys.json
new file mode 100644
index 0000000000000000000000000000000000000000..9748ea9411a1addce919ec47b75ad3f9805fae4e
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/hurdles_freestanding/trajectories/embodied-olympics-fun-hurdles-freestanding-i00-standard--wzyztys.json
@@ -0,0 +1,256 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-hurdles-freestanding-i00-standard--wzyztys",
+ "taskKey": "embodied-olympics-fun/hurdles_freestanding",
+ "dag_sha256": "631f06ede76d388c24511754cca206280db214cee4838b8a909aa9aebd7ae56e",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 1201,
+   "sim_steps": 1201,
+   "calls": 10,
+   "rows": 1201,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7251.2,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 4,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.517,
+   "peak_processes": 11,
+   "peak_disk_gb": 1.381,
+   "peak_cpus": 3.09,
+   "peak_pids": 107,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-hurdles-freestanding-i00-standard--wzyztys.mp4",
+  "duration_s": 12.04,
+  "fps": 25.0,
+  "frames": 301,
+  "sha256": "2141589006db2025d629ccf491315b0e011431d8936bd2da2f9acdbeba8a4ccc"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 5,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hurdle_1_cleared": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 12.0,
+      "log_step": 249,
+      "text_zh": "观察：视频 0–12.0 s 全景中机器人只沿跑道向前行走约数米，至视频 12.0 s 仍直立、位于第1栏（约 13.7 m）近端，未起跳；第1栏及后续各栏未被越过。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "避开",
+      "note_zh": "观察：视频 0–12 s 机器人双足行走接近第1栏，未与栏架接触，也未起跳。"
+     }
+    ]
+   },
+   "hurdle_2_cleared": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 12.0,
+      "log_step": 249,
+      "text_zh": "观察：第1栏尚未越过，前序节点未达成，视频 12.0 s 前机器人距本栏仍远。"
+     }
+    ]
+   },
+   "hurdle_3_cleared": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 12.0,
+      "log_step": 249,
+      "text_zh": "观察：第1栏尚未越过，前序节点未达成，视频 12.0 s 前机器人距本栏仍远。"
+     }
+    ]
+   },
+   "hurdle_4_cleared": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 12.0,
+      "log_step": 249,
+      "text_zh": "观察：第1栏尚未越过，前序节点未达成，视频 12.0 s 前机器人距本栏仍远。"
+     }
+    ]
+   },
+   "torso_at_finish": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 12.0,
+      "log_step": 249,
+      "text_zh": "观察：第1栏尚未越过，前序节点未达成，视频 12.0 s 前机器人距本栏仍远。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "time_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 140,
+   "direction": "lower_better"
+  },
+  {
+   "name": "distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": "higher_better"
+  },
+  {
+   "name": "hurdles_cleared",
+   "value": 0,
+   "unit": "count",
+   "threshold": 4,
+   "direction": "higher_better"
+  },
+  {
+   "name": "hurdles_down",
+   "value": null,
+   "unit": "count",
+   "threshold": null,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hurdle_1_cleared",
+   "where": "转场",
+   "t": 12.0,
+   "log_step": 249,
+   "evidence_zh": "观察：视频 12.04 s 的整段里机器人只在起点附近向栏架方向行走，未到第1栏即结束（评分器 first_reason_code 4，1201 个仿真步≈12 s）；没有起跳或越栏。"
+  },
+  "phenomena": [
+   {
+    "t": 12.0,
+    "log_step": 249,
+    "evidence_zh": "观察：视频 12.0 s 机器人仍直立，但行进距离远不足到达第1栏（约 13.7 m），行走过慢/过短；未见跌倒。",
+    "cat": "全身",
+    "item": "步态"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "根因未给出：无状态日志，log 中最后交付脚本 gen.py（log step 249）的运行输出未记录，不能判断 12 s 即终止的原因（agent 自述无）；log step 241 有 Gemini 过滤拦截，但之后 agent 仍继续运行并交付，不按外部因素处理。视频全景分辨率低，距离为估计。",
+  "unmet": {
+   "hurdle_1_cleared": false,
+   "hurdle_2_cleared": false,
+   "hurdle_3_cleared": false,
+   "hurdle_4_cleared": false,
+   "torso_at_finish": false
+  }
+ },
+ "summary_zh": "观察：graded 视频仅 12 s，机器人行走约数米后结束，未到第1栏，0/5 节点达成。断点为越过第1栏（转场阶段，未起跳），现象为步态（行进过短）；根因因证据不足留空。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频 12.04 s 全程机器人在第1栏前约 7 m 处直立行走，0/5 节点未达成、断点 hurdle_1_cleared、现象 步态（D25(c)，行进中被截止）均有帧证据支持；根因留空并附 gap_zh 可接受。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:59Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-hurdles-freestanding-i00-standard--wzyztys/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-hurdles-freestanding-i00-standard--wzyztys/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "eff2437a5b4e096813dbd4eadd78cbd992835f08c96e902a511966213e950fea",
+   "video_sha256": "2141589006db2025d629ccf491315b0e011431d8936bd2da2f9acdbeba8a4ccc",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/juggling_feet/trajectories/embodied-olympics-fun-juggling-feet-i00-standard--rbm4jr5.json b/phaseR/embodied-olympics-fun/juggling_feet/trajectories/embodied-olympics-fun-juggling-feet-i00-standard--rbm4jr5.json
new file mode 100644
index 0000000000000000000000000000000000000000..8956d25bc6ab6302069451a8ce280adbebf6ec73
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/juggling_feet/trajectories/embodied-olympics-fun-juggling-feet-i00-standard--rbm4jr5.json
@@ -0,0 +1,300 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-juggling-feet-i00-standard--rbm4jr5",
+ "taskKey": "embodied-olympics-fun/juggling_feet",
+ "dag_sha256": "6fb5f868dd0ec1e7ac4f8d9035cd8456789adc45e6c5ceda5fec1ad9d3ba10f6",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 0,
+   "sim_steps": 0,
+   "calls": 23,
+   "rows": 0,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7248.8,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 4,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.986,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.388,
+   "peak_cpus": 3.66,
+   "peak_pids": 109,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-juggling-feet-i00-standard--rbm4jr5.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "12d147d4a5f031b87ce4c65888a462965aa52f2b5691f991b4e3a59da41625a8"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 10,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "juggles_counted_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第1次颠球未达成。"
+     }
+    ]
+   },
+   "juggles_counted_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第2次颠球未达成。"
+     }
+    ]
+   },
+   "juggles_counted_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第3次颠球未达成。"
+     }
+    ]
+   },
+   "juggles_counted_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第4次颠球未达成。"
+     }
+    ]
+   },
+   "juggles_counted_5": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第5次颠球未达成。"
+     }
+    ]
+   },
+   "juggles_counted_6": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第6次颠球未达成。"
+     }
+    ]
+   },
+   "juggles_counted_7": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第7次颠球未达成。"
+     }
+    ]
+   },
+   "juggles_counted_8": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第8次颠球未达成。"
+     }
+    ]
+   },
+   "juggles_counted_9": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第9次颠球未达成。"
+     }
+    ]
+   },
+   "juggles_counted_10": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：回放视频仅1帧（0.04 s），球仍夹在机器人双掌之间，机器人静止站立，没有任何颠球；run.json reward 中 n_actions=0、sim_steps=0，log 中从未向机器人提交动作（sim.submit=0），没有触球发生。第10次颠球未达成。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "touches",
+   "value": null,
+   "unit": "count",
+   "threshold": 10,
+   "direction": "higher_better"
+  },
+  {
+   "name": "ball_height_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "juggles_counted_1",
+   "where": "首步前",
+   "t": 0.0,
+   "log_step": null,
+   "evidence_zh": "观察：graded 视频仅1帧，球仍在双掌之间；reward 中 n_actions=0、sim_steps=0，log 全程只在 agent 本地的 MuJoCo 仿真里调参，从未对真实环境发送动作（log step 3 后无提交，最后 step 210 仍在本地诊断）。第1次颠球从未开始。"
+  },
+  "phenomena": [
+   {
+    "t": 0.0,
+    "log_step": null,
+    "evidence_zh": "观察：graded 视频只有初始帧，机器人未动、球未被脚触碰；n_actions=0。",
+    "cat": "没尝试",
+    "item": "没尝试"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "根因层为空：只有“未交付任何动作”，不能由没尝试或缺失交付推出根因；agent 用约7200 s 在本地仿真里调试单腿颠球控制器而从未提交（log step 175-210），用完时间不是根因。视频仅1帧，无法观察更多。",
+  "unmet": {
+   "juggles_counted_1": false,
+   "juggles_counted_2": false,
+   "juggles_counted_3": false,
+   "juggles_counted_4": false,
+   "juggles_counted_5": false,
+   "juggles_counted_6": false,
+   "juggles_counted_7": false,
+   "juggles_counted_8": false,
+   "juggles_counted_9": false,
+   "juggles_counted_10": false
+  }
+ },
+ "summary_zh": "观察：graded 视频仅1帧，机器人静止、球在双掌间，n_actions=0，未发送任何动作；10个颠球节点均未达成（0/10）。断点为第1次颠球（首步前，没尝试）；agent 在 agent 工作约7200 s 内只做本地仿真调参（log step 2-210）而未交付，根因无证据，留空。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "单帧回放显示球仍在双掌间、机器人静止；log 全程 sim.submit=0，n_actions=0，10个节点未达成、断点首步前/没尝试、根因留空并写 gap_zh，均有证据支持。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:34Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-juggling-feet-i00-standard--rbm4jr5/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-juggling-feet-i00-standard--rbm4jr5/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "0be47505bb91b501d326dd2b1d85ce9c4004a460e3e2cb5c9b4d3af1dbb296d1",
+   "video_sha256": "12d147d4a5f031b87ce4c65888a462965aa52f2b5691f991b4e3a59da41625a8",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/layup/trajectories/embodied-olympics-fun-layup-i00-standard--n4vopej.json b/phaseR/embodied-olympics-fun/layup/trajectories/embodied-olympics-fun-layup-i00-standard--n4vopej.json
new file mode 100644
index 0000000000000000000000000000000000000000..935de4f84cc2853f0dd8f22f90c91db69d0b5cf6
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/layup/trajectories/embodied-olympics-fun-layup-i00-standard--n4vopej.json
@@ -0,0 +1,209 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-layup-i00-standard--n4vopej",
+ "taskKey": "embodied-olympics-fun/layup",
+ "dag_sha256": "cb7245d393b71eb13fe5757b27868d83a05141e8ffe5b75b752a5b5a686820f3",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 114,
+   "sim_steps": 114,
+   "calls": 6,
+   "rows": 114,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 5346.5,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.774,
+   "peak_processes": 10,
+   "peak_disk_gb": 1.542,
+   "peak_cpus": 2.04,
+   "peak_pids": 143,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-layup-i00-standard--n4vopej.mp4",
+  "duration_s": 1.16,
+  "fps": 25.0,
+  "frames": 29,
+  "sha256": "ab35ee4b6235470d4c79011fb76cf6a88462022f455c763d45ed6e1fab4f99e9"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "layup_scored": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.12,
+      "log_step": null,
+      "text_zh": "观察：视频 0–1.16 s（29 帧）中机器人始终双脚在地、双手持球向篮板方向走/跑，视频 1.12 s 仍未起跳、球未离手，球也未飞向篮圈；视频在此结束。未达到上篮命中。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "双臂协作",
+      "how": "抓握",
+      "note_zh": "观察：视频 0–1.16 s 机器人双手（Inspire 灵巧手）一直把篮球持在胸前，球未离手。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：双脚交替在地板上迈步支撑前进，未见腾空。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "layup_scored",
+   "t": 1.1,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 1.16 s 内机器人持球向前走，未起跳、未出手，视频在 1.16 s 结束。观察（基准服务读数，log step 193）：被评分 episode 在 sim 1.14 s、114 个动作行时以“travelling: a foot slid along the floor”判失败，stages_reached=0/3（run.json reward.n_actions=114，first_reason_code=2）。log step 266 的本地仿真复现（travelling_slide_left）仅作辅助。",
+   "log_step": 193
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "步态",
+    "t": 1.1,
+    "log_step": 193,
+    "evidence_zh": "观察：视频 0–1.16 s 机器人持球走步前进，始终未腾空起跳；观察（服务读数，log step 193）：被评分 episode 因“travelling: a foot slid along the floor”（脚滑移走步）判失败，stages_reached=0/3。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：agent 自写的开环 gen.py 轨迹在本地仿真与远端实际执行不一致（log step 245、266 中 loc_quat 与远端 rem_quat 有差异），脚滑移触发走步判罚；子类型不明，故只记控制。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "无状态日志，视频仅 1.16 s；终止原因已有服务读数（log step 193）。根因（开环轨迹在远端执行与本地仿真不一致导致脚滑移）为推断，子类型不明。",
+  "unmet": {
+   "layup_scored": false
+  }
+ },
+ "summary_zh": "观察：被评分的 episode 仅 1.16 s，机器人双手持球向篮板方向走步，未起跳、未出手，上篮命中未达成（0/1）。服务读数（log step 193）：判罚为走步（脚滑移），stages_reached 0/3。断点为上篮命中的助跑/起跳阶段，现象记为步态，根因仅能记控制（开环轨迹本地与远端不一致，推断）。主要缺口：无状态日志，视频仅 1.16 s。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点（未达成）、断点 layup_scored、现象 步态 的类别均与视频和服务读数一致；问题只在证据来源：终止原因（走步，脚滑移）在 log step 193 有基准服务对被评分 episode 的直接读数（outcome: travelling: a foot slid along the floor; stages_reached 0/3; 114 行），文件却把它写成只来自 agent 本地仿真的推断，并因此把 conf 压到 4、gap 写成缺证据。",
+    "n_issues": 3,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点 layup_scored(操作中)、现象步态均与视频 0–1.16 s(持球走步、未起跳)及 log step 193 服务读数(travelling,stages_reached 0)一致;第一轮提出的证据来源问题已修正,未发现新错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:07Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-layup-i00-standard--n4vopej/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-layup-i00-standard--n4vopej/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-layup-i00-standard--n4vopej/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-layup-i00-standard--n4vopej/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "78fe5b8761a366865a5600f4f5d641b18e45fe12392f76f863f2850c447504dc",
+   "video_sha256": "ab35ee4b6235470d4c79011fb76cf6a88462022f455c763d45ed6e1fab4f99e9",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/long_rope/trajectories/embodied-olympics-fun-long-rope-i00-privileged--jkqffee.json b/phaseR/embodied-olympics-fun/long_rope/trajectories/embodied-olympics-fun-long-rope-i00-privileged--jkqffee.json
new file mode 100644
index 0000000000000000000000000000000000000000..33e61bc8841cd111a965167fac6aec32f66bc52f
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/long_rope/trajectories/embodied-olympics-fun-long-rope-i00-privileged--jkqffee.json
@@ -0,0 +1,368 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-long-rope-i00-privileged--jkqffee",
+ "taskKey": "embodied-olympics-fun/long_rope",
+ "dag_sha256": "428ce1595ca927c3521a00d9a8ef26be24ff8e7f72d3d4261ef1c779bc1d5c12",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 1055,
+   "sim_steps": 1055,
+   "horizon": 2000,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 12,
+   "stages_total": 12,
+   "fell": 0,
+   "replay_s": 20.7,
+   "bad_trajectory": 0,
+   "metric_jumps": 10.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 35,
+   "agent_resets": 336,
+   "agent_replays": 5,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-long-rope-i00-privileged--jkqffee/replay.mp4",
+  "duration_s": 10.56,
+  "fps": 25.0,
+  "frames": 264,
+  "sha256": "ae26cbe5807029efa7b9f0ac628e912017514fd2ff2bd8ce0dd1e8ae93c96aed"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 10,
+  "base": 0,
+  "peak": 10,
+  "final": 10,
+  "nodes": {
+   "rope_cleared_1": {
+    "method": "visual",
+    "intervals": [
+     [
+      0.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 0.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第1次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   },
+   "rope_cleared_2": {
+    "method": "visual",
+    "intervals": [
+     [
+      1.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 1.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 1.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第2次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   },
+   "rope_cleared_3": {
+    "method": "visual",
+    "intervals": [
+     [
+      2.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 2.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 2.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第3次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   },
+   "rope_cleared_4": {
+    "method": "visual",
+    "intervals": [
+     [
+      3.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 3.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 3.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第4次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   },
+   "rope_cleared_5": {
+    "method": "visual",
+    "intervals": [
+     [
+      4.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 4.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 4.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第5次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   },
+   "rope_cleared_6": {
+    "method": "visual",
+    "intervals": [
+     [
+      5.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 5.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 5.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第6次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   },
+   "rope_cleared_7": {
+    "method": "visual",
+    "intervals": [
+     [
+      6.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 6.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 6.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第7次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   },
+   "rope_cleared_8": {
+    "method": "visual",
+    "intervals": [
+     [
+      7.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 7.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 7.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第8次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   },
+   "rope_cleared_9": {
+    "method": "visual",
+    "intervals": [
+     [
+      8.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 8.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 8.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第9次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   },
+   "rope_cleared_10": {
+    "method": "visual",
+    "intervals": [
+     [
+      9.85,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 9.85,
+      "log_step": null,
+      "text_zh": "观察：视频约 9.85 s 附近机器人双脚离地腾空、细绳从脚下通过（按每0.17 s一格的裁剪图估计，每跳约间隔1.0 s，时间误差约±0.2 s）；评分器 metric_jumps=10、stages 12/12、success=1，第10次过绳计入。推断：具体计数时刻由等间隔腾空动作估计，未逐毫秒核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在木地板上蹬地起跳并落地；绳未触身（评分器 fell=0、成功）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "jumps",
+   "value": 10,
+   "unit": "次",
+   "threshold": 10,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：评分器 metric_jumps=10、stages 12/12、fell=0，10个过绳节点均达成。视频 10.56 s，约每1.0 s一次腾空（视频约 0.85–9.9 s），各节点时间由裁剪图估计，无状态日志，精确过绳时刻未逐毫秒核对。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "成功运行，10个过绳节点均为真，final=10 与评分器 metric_jumps=10、stages 12/12 一致；视频约每1.0 s一次腾空，节点时间（0.85–9.85 s）与接触标注合理，无失败层，无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:46Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-long-rope-i00-privileged--jkqffee/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-long-rope-i00-privileged--jkqffee/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "88e5c01ea84d5454da0a124861edb7049a1050da95d42129c5cd5b1e57f77837",
+   "video_sha256": "ae26cbe5807029efa7b9f0ac628e912017514fd2ff2bd8ce0dd1e8ae93c96aed",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/paddle_bounce/trajectories/embodied-olympics-fun-paddle-bounce-i00-realtime--pmyatk3.json b/phaseR/embodied-olympics-fun/paddle_bounce/trajectories/embodied-olympics-fun-paddle-bounce-i00-realtime--pmyatk3.json
new file mode 100644
index 0000000000000000000000000000000000000000..3e531449605ef9d5e2b3fa034e54be0c4bcf66b7
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/paddle_bounce/trajectories/embodied-olympics-fun-paddle-bounce-i00-realtime--pmyatk3.json
@@ -0,0 +1,455 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-paddle-bounce-i00-realtime--pmyatk3",
+ "taskKey": "embodied-olympics-fun/paddle_bounce",
+ "dag_sha256": "0430fbd3fb280dd4add17cdc2f794688f20fdd9ca4d316d2240c5df644bbea14",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 252,
+   "sim_steps": 252,
+   "calls": 171,
+   "rows": 1059,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 6067.7,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "realtime": 1,
+   "slowdown": 1.0,
+   "clock_s": 3.771,
+   "max_lag_s": 0.5051,
+   "stalls": 2,
+   "stalled_s": 1.009,
+   "stalled_share": 0.2677,
+   "held_steps": 0,
+   "late_steps": 17,
+   "requests_refused": 2,
+   "request_busy_s": 0.69,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.933,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.382,
+   "peak_cpus": 3.93,
+   "peak_pids": 223,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-paddle-bounce-i00-realtime--pmyatk3.mp4",
+  "duration_s": 2.56,
+  "fps": 25.0,
+  "frames": 64,
+  "sha256": "af1783595a79dbf837a085cbf2a304f26f8a6ba9d6db14021db38e5fee008dbf"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 20,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "bounce_1": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "其他",
+      "note_zh": "观察：右手持拍，拍面未触球；球释放后视频结束前未见球拍与球接触。"
+     }
+    ]
+   },
+   "bounce_2": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_3": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_4": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_5": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_6": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_7": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_8": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_9": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_10": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_11": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_12": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_13": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_14": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_15": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_16": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_17": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_18": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_19": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   },
+   "bounce_20": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 2.5,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.56 s，机器人右手持红色球拍从胸前下垂到身体侧下方（约 0.5-1.5 s），拍面朝下偏向身侧；约 2.0 s 杯中橙球释放，视频 2.56 s 结束前球未被拍面击起，无任何一次击球。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "bounces",
+   "value": 0,
+   "unit": "count",
+   "threshold": 20,
+   "direction": "higher_better"
+  },
+  {
+   "name": "ball_height_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 1.2,
+   "direction": "higher_better"
+  },
+  {
+   "name": "grip_break_max_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "bounce_1",
+   "t": 1.0,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.5-1.5 s 机器人把拍放低到身侧，视频 2.0 s 球释放后直到 2.56 s 结束拍面未在球下方，未击到球。",
+   "log_step": null
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "t": 2.5,
+    "log_step": null,
+    "evidence_zh": "观察：视频 1-2.5 s 拍面低垂于髋侧、朝下倾斜，球在身体前上方落下，拍子未被带到球下方。",
+    "item": "位姿不准"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "视频仅 2.56 s，且为实时模式，无状态日志；拍面偏离的具体原因（控制脚本/估计）无足够证据，故根因留空。最后几步 agent 在调试球检测与落点估计（log step 273-278），属会话证据，不作根因。",
+  "unmet": {
+   "bounce_1": false,
+   "bounce_2": false,
+   "bounce_3": false,
+   "bounce_4": false,
+   "bounce_5": false,
+   "bounce_6": false,
+   "bounce_7": false,
+   "bounce_8": false,
+   "bounce_9": false,
+   "bounce_10": false,
+   "bounce_11": false,
+   "bounce_12": false,
+   "bounce_13": false,
+   "bounce_14": false,
+   "bounce_15": false,
+   "bounce_16": false,
+   "bounce_17": false,
+   "bounce_18": false,
+   "bounce_19": false,
+   "bounce_20": false
+  }
+ },
+ "summary_zh": "观察：视频 2.56 s 内右手持拍被放低到身侧，球释放后未被击中，20 个颠球节点均未达成（0/20）。断点为第1次颠球，现象为位姿不准；视频极短且无状态日志，根因无法确定。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "视频仅2.56 s：右手持拍0.5-1.5 s放低到身侧，约2.0 s球释放后下落，拍面未在球下方，无击球；20个节点均未达成、断点bounce_1、位姿不准与证据一致，根因留空并附gap_zh合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:42Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-paddle-bounce-i00-realtime--pmyatk3/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-paddle-bounce-i00-realtime--pmyatk3/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "01459a77bacea3ea4db8e359a8483617d29a5e847bc6cf5887fe04b988338e97",
+   "video_sha256": "af1783595a79dbf837a085cbf2a304f26f8a6ba9d6db14021db38e5fee008dbf",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/pitch_target/trajectories/embodied-olympics-fun-pitch-target-i00-standard--yywpbtb.json b/phaseR/embodied-olympics-fun/pitch_target/trajectories/embodied-olympics-fun-pitch-target-i00-standard--yywpbtb.json
new file mode 100644
index 0000000000000000000000000000000000000000..54e0b80fa94f171aea8db8c0e7f1d6cb63d886b9
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/pitch_target/trajectories/embodied-olympics-fun-pitch-target-i00-standard--yywpbtb.json
@@ -0,0 +1,187 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-pitch-target-i00-standard--yywpbtb",
+ "taskKey": "embodied-olympics-fun/pitch_target",
+ "dag_sha256": "b9c9d2241d67e603a33c144ea507b4d070461d1a518b4397faeaecf97bb7bb6f",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 239,
+   "sim_steps": 239,
+   "calls": 5,
+   "rows": 239,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 3848.3,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.157,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.326,
+   "peak_cpus": 1.98,
+   "peak_pids": 146,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-pitch-target-i00-standard--yywpbtb.mp4",
+  "duration_s": 2.4,
+  "fps": 25.0,
+  "frames": 60,
+  "sha256": "3f226903180af32cf6b586a08a27b1116c27dfae3fff81c92161bdc62d039684"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "pitch_strike": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.2,
+      "log_step": 127,
+      "text_zh": "观察：视频 1.0-1.2 s 右手持球的右臂开始前摆，白球在约 1.1 s 离手，只飞出很短距离即落在投手丘脚边，未飞向本垒；视频 2.4 s 内球再未出现在远处。服务状态读数（log step 127）outcome“the ball touched the ground before home plate”，reason failed，stages 1/2，与视频一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "右手五指持球，挥臂时抓握不住，球在仿真 0.98 s（log step 128 holding_right 1→0）早早脱手。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "双脚站在投手丘/投手板上，视频中未跌倒。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "pitch_strike",
+   "where": "操作中",
+   "t": 1.1,
+   "log_step": 127,
+   "evidence_zh": "观察：视频约 1.1 s 球从手中脱出，低速落在投手丘附近；log step 127 服务读数 outcome 为球在本垒前触地，stages 1/2。该球落地结束了本次评分。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 1.1,
+    "log_step": 128,
+    "evidence_zh": "观察：log step 128 的本体读数显示挥臂第2步（仿真 0.98 s）holding_right 由1变0，手指闭合度约0.74，球在手臂速度尚未建立时就离手；视频 1.1 s 前后球向下落到丘上，速度很小。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：挥臂开始即失去抓握，球以约1 m/s向下离手，释放时刻和速度均不符合设计；agent 在 log step 236-246 自己也在排查指尖位置与手-球平衡，说明手指抓握/挥臂加速度设计未能保持球直到合适释放点。具体子类不明。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "远景视频中球很小，仅能确认其离手早且很快落地；根因子类（抓取设计还是挥臂轨迹）无直接证据。",
+  "unmet": {
+   "pitch_strike": false
+  }
+ },
+ "summary_zh": "观察：视频 2.4 s 内机器人右手持球开始挥臂，球在约视频 1.1 s 过早脱手并在本垒前很近处落地（服务读数：球在本垒前触地，stages 1/2），未投出好球。断点为投出好球（操作中），现象为没保持住，根因归为控制（子类不明）。证据缺口：球在远景中很小。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与证据一致：视频总长 2.4 s 对应仿真 2.39 s（239 步），球在仿真约 0.98 s 随挥臂脱手并在本垒前触地（服务读数 stages 1/2），断点 pitch_strike 操作中、现象没保持住、根因控制（子类不明）均有 log 128 的 holding_right 1→0 支撑，未见需改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:57:01Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-pitch-target-i00-standard--yywpbtb/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-pitch-target-i00-standard--yywpbtb/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "1eb6fcfe8c48824d59f87c26ce04492f975b241285fd04ee46f4e8d0b8fad3de",
+   "video_sha256": "3f226903180af32cf6b586a08a27b1116c27dfae3fff81c92161bdc62d039684",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/platform_10m/trajectories/embodied-olympics-fun-platform-10m-i00-privileged--dftqx9u.json b/phaseR/embodied-olympics-fun/platform_10m/trajectories/embodied-olympics-fun-platform-10m-i00-privileged--dftqx9u.json
new file mode 100644
index 0000000000000000000000000000000000000000..5d424073cac9a1ccad602d3adfd3099459dfc99c
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/platform_10m/trajectories/embodied-olympics-fun-platform-10m-i00-privileged--dftqx9u.json
@@ -0,0 +1,237 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-platform-10m-i00-privileged--dftqx9u",
+ "taskKey": "embodied-olympics-fun/platform_10m",
+ "dag_sha256": "aa4b50e0889b1128e2a630e2ad361b2232bedadcd3de6b1ecbe04bf0545d6da2",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 233,
+   "sim_steps": 233,
+   "horizon": 2000,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 3,
+   "stages_total": 3,
+   "fell": 0,
+   "replay_s": 3.05,
+   "bad_trajectory": 0,
+   "metric_entry_apart": 0.1289,
+   "metric_entry_beyond": 0.3574,
+   "metric_entry_distance_m": 2.221,
+   "metric_entry_feet": 0.0944,
+   "metric_entry_hip": 171.7174,
+   "metric_entry_knee": 170.1274,
+   "metric_entry_off": 0.0655,
+   "metric_entry_tilt_deg": 0.1,
+   "metric_entry_toes": 0.45,
+   "metric_entry_twist": 0.66,
+   "metric_somersault_deg": 174.8,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 13,
+   "agent_resets": 5,
+   "agent_replays": 7,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.717,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.25,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-platform-10m-i00-privileged--dftqx9u/replay.mp4",
+  "duration_s": 2.36,
+  "fps": 25.0,
+  "frames": 59,
+  "sha256": "b702811016209a8f09064a054e0e0458abb2d6467c808ac30a6a8e4c80301fe2"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "forward_dive_completed": {
+    "method": "visual",
+    "intervals": [
+     [
+      2.25,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 2.25,
+      "log_step": 73,
+      "text_zh": "观察：视频 1.6 s 机器人在空中近水平翻转；2.0-2.2 s 身体近竖直、头手朝下穿过水面；2.3 s 起画面中已不见机器人（全身没入）。评分器 success=1、stages 3/3、fell=0。推断：完成时刻约视频 2.25 s（机器人很小，时间为估计）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频起始阶段机器人双脚在十米台前沿蹬离（画面中机器人极小，细节难辨）；腾空后无外部接触，入水为终点。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "somersault_deg",
+   "value": 174.8,
+   "unit": "deg",
+   "threshold": 180,
+   "direction": "target"
+  },
+  {
+   "name": "entry_distance_m",
+   "value": 2.221,
+   "unit": "m",
+   "threshold": [
+    0.5,
+    3.0
+   ],
+   "direction": "target"
+  },
+  {
+   "name": "entry_tilt_deg",
+   "value": 0.1,
+   "unit": "deg",
+   "threshold": 15,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_hip",
+   "value": 171.7174,
+   "unit": "deg",
+   "threshold": 160,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_knee",
+   "value": 170.1274,
+   "unit": "deg",
+   "threshold": 160,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_feet",
+   "value": 0.0944,
+   "unit": "m",
+   "threshold": 0.12,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_toes",
+   "value": 0.45,
+   "unit": "rad",
+   "threshold": 0.3,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_beyond",
+   "value": 0.3574,
+   "unit": "m",
+   "threshold": 0.15,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_apart",
+   "value": 0.1289,
+   "unit": "m",
+   "threshold": 0.2,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_off",
+   "value": 0.0655,
+   "unit": "m",
+   "threshold": 0.15,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_twist",
+   "value": 0.66,
+   "unit": "deg",
+   "threshold": 20,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：视频 2.36 s 内机器人从十米台蹬离，翻转后在约视频 2.25 s 近竖直入水并全身没入，评分器 success=1，stages 3/3，未摔倒。无状态日志，机器人在侧视画面中很小，节点时间为视觉估计。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "成功运行，单节点 forward_dive_completed 达成、final=true 与评分器 success=1、stages 3/3 一致；约 2.25 s 完成时间在视频 2.36 s 内合理，无失败层需标注。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:25Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-platform-10m-i00-privileged--dftqx9u/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-platform-10m-i00-privileged--dftqx9u/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "1e0d23e050cf88b5c03c28482c7abfde0f8b379e9c9e19737b4f5337f2843983",
+   "video_sha256": "b702811016209a8f09064a054e0e0458abb2d6467c808ac30a6a8e4c80301fe2",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/platform_10m/trajectories/embodied-olympics-fun-platform-10m-i00-privileged--xbuutg2.json b/phaseR/embodied-olympics-fun/platform_10m/trajectories/embodied-olympics-fun-platform-10m-i00-privileged--xbuutg2.json
new file mode 100644
index 0000000000000000000000000000000000000000..1422b0280a2d374809ae54d3011f651da48797cb
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/platform_10m/trajectories/embodied-olympics-fun-platform-10m-i00-privileged--xbuutg2.json
@@ -0,0 +1,235 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-platform-10m-i00-privileged--xbuutg2",
+ "taskKey": "embodied-olympics-fun/platform_10m",
+ "dag_sha256": "aa4b50e0889b1128e2a630e2ad361b2232bedadcd3de6b1ecbe04bf0545d6da2",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 251,
+   "sim_steps": 251,
+   "horizon": 2000,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 3,
+   "stages_total": 3,
+   "fell": 0,
+   "replay_s": 3.49,
+   "bad_trajectory": 0,
+   "metric_entry_apart": 0.1339,
+   "metric_entry_beyond": 0.3713,
+   "metric_entry_distance_m": 2.231,
+   "metric_entry_feet": 0.0851,
+   "metric_entry_hip": 172.118,
+   "metric_entry_knee": 170.0833,
+   "metric_entry_off": 0.067,
+   "metric_entry_tilt_deg": 0.1,
+   "metric_entry_toes": 0.5001,
+   "metric_entry_twist": 0.1703,
+   "metric_somersault_deg": 175.3,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 60,
+   "agent_resets": 18259,
+   "agent_replays": 6,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-platform-10m-i00-privileged--xbuutg2/replay.mp4",
+  "duration_s": 2.52,
+  "fps": 25.0,
+  "frames": 63,
+  "sha256": "26398d73336e145f4a2bfc78ad6b2886896283f5040da898cc2de2b99c3a61ae"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "forward_dive_completed": {
+    "method": "visual",
+    "intervals": [
+     [
+      2.4,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 2.4,
+      "log_step": null,
+      "text_zh": "观察：视频 0–1.5 s 机器人在台前沿起跳后腾空翻转（画面中机器人很小）；视频约 2.2–2.4 s 身体近竖直入水，最后帧在水面下。评分器字段：success=1，somersault_deg 175.3，entry_distance_m 2.231，entry_tilt_deg 0.1，fell=0，stages 3/3。推断：入水完成时刻约视频 2.4 s（机位远，精度约 0.2 s）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在十米台前沿蹬离，随后腾空无外部接触（机位远，仅能看到整体姿态）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "somersault_deg",
+   "value": 175.3,
+   "unit": "deg",
+   "threshold": 180,
+   "direction": "target"
+  },
+  {
+   "name": "entry_distance_m",
+   "value": 2.231,
+   "unit": "m",
+   "threshold": [
+    0.5,
+    3.0
+   ],
+   "direction": "target"
+  },
+  {
+   "name": "entry_tilt_deg",
+   "value": 0.1,
+   "unit": "deg",
+   "threshold": 15,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_hip",
+   "value": 172.118,
+   "unit": "deg",
+   "threshold": 160,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_knee",
+   "value": 170.0833,
+   "unit": "deg",
+   "threshold": 160,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_feet",
+   "value": 0.0851,
+   "unit": "m",
+   "threshold": 0.12,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_toes",
+   "value": 0.5001,
+   "unit": "rad",
+   "threshold": 0.3,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_beyond",
+   "value": 0.3713,
+   "unit": "m",
+   "threshold": 0.15,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_apart",
+   "value": 0.1339,
+   "unit": "m",
+   "threshold": 0.2,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_off",
+   "value": 0.067,
+   "unit": "m",
+   "threshold": 0.15,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_twist",
+   "value": 0.1703,
+   "unit": "deg",
+   "threshold": 20,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：机器人从十米台起跳，翻转约175.3°，约视频 2.4 s 近竖直入水，评分器 success=1、无违规。缺口：机位远，机器人很小，具体姿态细节只靠评分器字段。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "成功运行，唯一节点 forward_dive_completed 在视频约 2.4 s 入水，与帧和评分器字段（success=1，stages 3/3）一致；contact_actual 合理，无失败层。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:31Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-platform-10m-i00-privileged--xbuutg2/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-platform-10m-i00-privileged--xbuutg2/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "8054296250dc0e63610ce8a2eef85216f8d00d94ffcea3e22b722ccbf17bbee5",
+   "video_sha256": "26398d73336e145f4a2bfc78ad6b2886896283f5040da898cc2de2b99c3a61ae",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/platform_10m/trajectories/embodied-olympics-fun-platform-10m-i00-standard--ob8vvzg.json b/phaseR/embodied-olympics-fun/platform_10m/trajectories/embodied-olympics-fun-platform-10m-i00-standard--ob8vvzg.json
new file mode 100644
index 0000000000000000000000000000000000000000..2eb2b42e4ebf3699da086a7bf9f3f7dc3a267171
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/platform_10m/trajectories/embodied-olympics-fun-platform-10m-i00-standard--ob8vvzg.json
@@ -0,0 +1,248 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-platform-10m-i00-standard--ob8vvzg",
+ "taskKey": "embodied-olympics-fun/platform_10m",
+ "dag_sha256": "aa4b50e0889b1128e2a630e2ad361b2232bedadcd3de6b1ecbe04bf0545d6da2",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 2,
+   "sim_steps": 2,
+   "calls": 5,
+   "rows": 2,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 2625.0,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 4,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.815,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.316,
+   "peak_cpus": 1.68,
+   "peak_pids": 77,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-platform-10m-i00-standard--ob8vvzg.mp4",
+  "duration_s": 0.04,
+  "fps": 25.0,
+  "frames": 1,
+  "sha256": "04584f8f60456a274701be1106c3a461573a00261fa8463992737b36219e22ec"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "forward_dive_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.0,
+      "log_step": null,
+      "text_zh": "观察：graded 视频仅 1 帧（视频 0.00 s，0.04 s 长），只见十米台塔体、水面与看台，画面中未见机器人腾空或入水；run.json 记 n_actions=2、sim_steps=2。推断：被评分的回合只运行了 2 个控制步，没有起跳、翻转或入水，节点未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "不适用",
+      "how": "不适用",
+      "note_zh": "评分回合没有交付轨迹（只有 2 个控制步，agent 之后在本地 MuJoCo 中调参，log step 63-77），无可标注的接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "somersault_deg",
+   "value": null,
+   "unit": "deg",
+   "threshold": 180,
+   "direction": "target"
+  },
+  {
+   "name": "entry_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": [
+    0.5,
+    3.0
+   ],
+   "direction": "target"
+  },
+  {
+   "name": "entry_tilt_deg",
+   "value": null,
+   "unit": "deg",
+   "threshold": 15,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_hip",
+   "value": null,
+   "unit": "deg",
+   "threshold": 160,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_knee",
+   "value": null,
+   "unit": "deg",
+   "threshold": 160,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_feet",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.12,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_toes",
+   "value": null,
+   "unit": "rad",
+   "threshold": 0.3,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_beyond",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.15,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_apart",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.2,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_off",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.15,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_twist",
+   "value": null,
+   "unit": "deg",
+   "threshold": 20,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "unknown",
+   "t": null,
+   "log_step": 78,
+   "where": "首步前",
+   "evidence_zh": "观察：log step 78、80、82、84 的 agent 输出均为“This request was blocked by Gemini's filters”（模型提供方过滤器拦截），此后没有任何提交；评分回合仅 2 个控制步、视频 1 帧。推断：外部因素在交付前切断了 run，未交付跳水轨迹。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "外部因素：agent 在 log step 78 起连续被 Gemini 提供方过滤器拦截（log step 78-84），尚未交付任何跳水轨迹，故不给现象和根因（不是没尝试）。视频只有 1 帧，无法看到机器人动作。agent 在 log step 63-76 的本地仿真只是会话内测试，不作为现象。",
+  "unmet": {
+   "forward_dive_completed": false
+  }
+ },
+ "summary_zh": "观察：评分回合仅 2 个控制步、视频 1 帧（0.04 s），机器人未起跳，“直体前翻入水”未达成（0/1）。agent 在本地 MuJoCo 中调参，log step 78-84 起被 Gemini 提供方过滤器拦截，未交付轨迹；属外部因素切断，不做现象与根因标注。证据缺口：视频过短，无法看到机器人。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "评分回合仅 2 个控制步、视频 1 帧（只见塔体，无机器人腾空），节点未达成；log step 78-84 连续被 Gemini 提供方过滤器拦截，外部因素在交付前切断，空现象/根因并附 gap_zh 的处理正确。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:46Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-platform-10m-i00-standard--ob8vvzg/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-platform-10m-i00-standard--ob8vvzg/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "338d88908f05941d912951f7398d477e42246ff6cf2300e3d2da4e706514e50d",
+   "video_sha256": "04584f8f60456a274701be1106c3a461573a00261fa8463992737b36219e22ec",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/pole_vault/trajectories/embodied-olympics-fun-pole-vault-i00-standard--ntknrjl.json b/phaseR/embodied-olympics-fun/pole_vault/trajectories/embodied-olympics-fun-pole-vault-i00-standard--ntknrjl.json
new file mode 100644
index 0000000000000000000000000000000000000000..41510b2201e27d0b82bc4eaa4b0b6c391318e1fc
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/pole_vault/trajectories/embodied-olympics-fun-pole-vault-i00-standard--ntknrjl.json
@@ -0,0 +1,176 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-pole-vault-i00-standard--ntknrjl",
+ "taskKey": "embodied-olympics-fun/pole_vault",
+ "dag_sha256": "563d65428d5344656f3de560235015c833808d9cce8bab83d845705ceab114c3",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 433,
+   "sim_steps": 1954,
+   "calls": 14,
+   "rows": 1954,
+   "resets": 3,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7119.5,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 3,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.195,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.349,
+   "peak_cpus": 6.41,
+   "peak_pids": 140,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_cpus": 1
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-pole-vault-i00-standard--ntknrjl.mp4",
+  "duration_s": 4.36,
+  "fps": 25.0,
+  "frames": 109,
+  "sha256": "9a3dbb7056e2ba81a422c7a4f91238bfbde43ec47192645ddfbddc36b89fd35a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "vault_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 4.3,
+      "log_step": null,
+      "text_zh": "观察：视频 0-4.36 s 固定场地相机始终只见跑道末端、深色插斗、蓝垫和完好的红白横杆，机器人全程不在画面内；未见插杆、腾空、过杆或落地，视频 4.3 s 横杆仍在立柱上。推断：撑杆越杆落地未达成（与评分器 success=0 一致）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "视频中机器人不可见，无法判断手/脚实际接触；日志无状态记录。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "support_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 0.15,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "vault_completed",
+   "t": null,
+   "log_step": 215,
+   "where": "未知",
+   "evidence_zh": "观察：视频 0-4.36 s 机器人不在画面内，没有插杆/过杆画面，横杆未动；agent 在 log step 215 交付 gen.py 并自述已实现（仅为其自述）。推断：未完成撑杆越杆，具体断在助跑、插杆还是腾空无法从视频确定。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "视频固定相机从未拍到机器人（开局在零线前 20 米，视频仅 4.36 s），无状态日志，故无法确定可见失败现象和根因；agent 会话中的测试（如 log step 195-198 松杆、骨盆姿态剧变）属会话证据，不作为现象。standard 模式下 agent 自述的实现不能代替画面。",
+  "unmet": {
+   "vault_completed": false
+  }
+ },
+ "summary_zh": "观察：视频 0-4.36 s 只拍到插斗、蓝垫和完好的横杆，机器人不在画面内，唯一节点“撑杆越杆落地”判为未达成（final false）。断点定在该步，但因无状态日志且机器人不可见，现象与根因均留空；证据缺口即机器人全程不可见。",
+ "conf": 4,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "视频 0-4.36 s 固定场地相机只见插斗、蓝垫和完好横杆，机器人始终不在画面内，唯一节点 final=false（与 success=0 一致）、断点 ref=vault_completed、where=未知、现象与根因留空并写明 gap_zh，均有证据支持；agent 自述未用作现象。无状态日志，无法进一步判定，未发现需要修改的内容。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:56Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-pole-vault-i00-standard--ntknrjl/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-pole-vault-i00-standard--ntknrjl/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "7301cf1120f786efbc0b123e1ca127f9db2149fec4fa1a294b5f67d26ff8b474",
+   "video_sha256": "9a3dbb7056e2ba81a422c7a4f91238bfbde43ec47192645ddfbddc36b89fd35a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/rugby_conversion/trajectories/embodied-olympics-fun-rugby-conversion-i00-standard--evi7sts.json b/phaseR/embodied-olympics-fun/rugby_conversion/trajectories/embodied-olympics-fun-rugby-conversion-i00-standard--evi7sts.json
new file mode 100644
index 0000000000000000000000000000000000000000..c0481c86d031477abc85808260093f552b5e919b
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/rugby_conversion/trajectories/embodied-olympics-fun-rugby-conversion-i00-standard--evi7sts.json
@@ -0,0 +1,188 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-rugby-conversion-i00-standard--evi7sts",
+ "taskKey": "embodied-olympics-fun/rugby_conversion",
+ "dag_sha256": "081cffc4f4d695f779d7796359dd6e3941fd545401eb6e3eecaa88fb6616cb9b",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 83,
+   "sim_steps": 140,
+   "calls": 7,
+   "rows": 140,
+   "resets": 2,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 6621.0,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.447,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.403,
+   "peak_cpus": 1.83,
+   "peak_pids": 202,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-rugby-conversion-i00-standard--evi7sts.mp4",
+  "duration_s": 0.84,
+  "fps": 25.0,
+  "frames": 21,
+  "sha256": "dacc0c5ca7b6044dd51767d113556064fecfb5200c3a239d9eb0f7ec0eca3d12"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "conversion_scored": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 0.8,
+      "log_step": 251,
+      "text_zh": "观察：视频 0.4-0.84 s 机器人右腿摆动踢球，末帧（视频 0.8 s）球低空飞行、尚未到门柱处，球高明显低于横梁，未见越线；观察：log step 251 服务返回 outcome“the ball touched the ground before it scored”，stages_reached 1/2（该次即视频中的图像）。推断：球落地于得分前，未得分。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "观察：机器人右脚踢击球座上的球（击打，候选名“击打”），视频约 0.4-0.5 s；球随后离座低飞。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：踢球时左脚支撑草地，踢后身体后仰倾斜。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "conversion_scored",
+   "t": 0.6,
+   "log_step": 251,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0.45-0.84 s 一脚踢出后球低、短飞行，末帧远低于横梁高度；log step 251 服务判定“球在得分前触地”（该规则终止了回合）。"
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 0.8,
+    "log_step": 251,
+    "evidence_zh": "观察：视频末帧球仍在低空、离门柱与横梁很远，击球后弹道过低过短，未达到 3 m 越线高度；服务返回球在得分前触地。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "alt_feasible": "未知",
+    "evidence_zh": "观察：log step 266-275 显示 agent 在仿真里反复扫参数（脚踝时机、q10_pre、dz）追求单次击球与足够 vz，最终交付的一脚弹道仍不足；击球方式（脚踝/膝摆动与击球点）未能产生足够上扬速度。推断：踢击控制设计不足，子类型不明。",
+    "source_zh": "日志 / 视频"
+   }
+  ],
+  "gap_zh": "无状态日志；视频仅 0.84 s，球落地的确切时刻与球座接触细节未能在画面中看清，球飞行高度由单一固定视角估计。",
+  "unmet": {
+   "conversion_scored": false
+  }
+ },
+ "summary_zh": "观察：唯一节点“踢球过门”未达成（final false）。机器人在视频约 0.45 s 一脚踢出，球低短飞行，服务判定球得分前触地（log step 251）。推断：踢击弹道不足（控制），视频仅 0.84 s，证据有限。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "单节点未达成、断点 conversion_scored/操作中、接触（腿/躯干 其他击打 + 支撑）与视频一致：视频约 0.5 s 踢出，球贴地低飞、未过横梁高度，之后机器人后仰倒地；log step 251 服务返回球在得分前触地。位姿不准与控制根因证据有限但不矛盾，未发现需要改动的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:57:00Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-rugby-conversion-i00-standard--evi7sts/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-rugby-conversion-i00-standard--evi7sts/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f54b809a3320344cc8cf9070ea8db2cf0461134717472110f48b9bb8e7409b3c",
+   "video_sha256": "dacc0c5ca7b6044dd51767d113556064fecfb5200c3a239d9eb0f7ec0eca3d12",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/shot_put/trajectories/embodied-olympics-fun-shot-put-i00-privileged--wrpuctl.json b/phaseR/embodied-olympics-fun/shot_put/trajectories/embodied-olympics-fun-shot-put-i00-privileged--wrpuctl.json
new file mode 100644
index 0000000000000000000000000000000000000000..9ef98198cf170a17d4b16eeda731edf2996d58c7
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/shot_put/trajectories/embodied-olympics-fun-shot-put-i00-privileged--wrpuctl.json
@@ -0,0 +1,193 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-shot-put-i00-privileged--wrpuctl",
+ "taskKey": "embodied-olympics-fun/shot_put",
+ "dag_sha256": "b330d54bdc4fab556e2c99f97d32becd13070b5a7fcba4a435eb7b97d8662afe",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 1609,
+   "sim_steps": 1609,
+   "horizon": 6000,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 3,
+   "stages_total": 3,
+   "fell": 0,
+   "replay_s": 19.05,
+   "bad_trajectory": 0,
+   "metric_distance_m": 6.664,
+   "metric_release_height_m": 1.952,
+   "metric_release_push_fraction": 0.926,
+   "metric_release_speed_m_s": 6.99,
+   "metric_release_time_s": 10.212,
+   "metric_stance_neck_distance_m": 0.159,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 159,
+   "agent_resets": 20016,
+   "agent_replays": 24,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-shot-put-i00-privileged--wrpuctl/replay.mp4",
+  "duration_s": 16.12,
+  "fps": 25.0,
+  "frames": 403,
+  "sha256": "f7b57623e523cdc778cc90a882d12104b0da5772af4801c6e0ebfd62fab5c629"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 2,
+  "final": 2,
+  "nodes": {
+   "shot_landed_valid": {
+    "method": "visual",
+    "intervals": [
+     [
+      11.7,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 10.2,
+      "log_step": 66,
+      "text_zh": "观察：视频 10.2 s 起右臂上举、铅球离手；评分器字段 release_time_s=10.212、distance_m=6.664（≥6.0）、stages 含 release、landing。推断：球于视频约 11.5-12 s 落地（视频 11.7 s 田野中可见小黑点，落点本身画面中不清晰），成绩由评分器认定有效。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "托/舀",
+      "note_zh": "右手无指掌垫从下方托住并推出铅球（视频约 9-10.2 s）"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "双脚支撑投掷圈并发力，颈侧躯干贴靠球"
+     }
+    ]
+   },
+   "rear_exit_standing": {
+    "method": "visual",
+    "intervals": [
+     [
+      16.0,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 16.0,
+      "log_step": 67,
+      "text_zh": "观察：视频 12-16.1 s 机器人在圈内向后移动并保持直立，视频末帧仍站立未倒；评分器 success=1、stages_reached 3/3、fell=0。agent 自述 16.15 s 站稳（log step 67），与视频时长 16.12 s 基本一致。推断：站稳在视频末尾达成，具体时刻约 16 s，不确定。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "双脚在圈内及后半部支撑，最后站稳"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "distance_m",
+   "value": 6.664,
+   "unit": "m",
+   "threshold": 6.0,
+   "direction": "higher_better"
+  },
+  {
+   "name": "release_push_fraction",
+   "value": 0.926,
+   "unit": "1",
+   "threshold": 0.5,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：右手在视频约 10.2 s 推出铅球，评分器记录成绩 6.664 m，并完成后半圈退场站稳（3/3 阶段，未摔倒）。落地点与站稳的精确时刻在单一远景视频中不清晰，时间为推断。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "成功运行，两节点 final=true 与评分器 3/3 阶段、success=1、release_time 10.212 s（约等于视频 10.2 s）一致；时间标注为推断，接触标签合理。未发现需修改的错误。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:41Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-shot-put-i00-privileged--wrpuctl/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-shot-put-i00-privileged--wrpuctl/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "6b09320a8d1441223ea1d7e50c0f60bd5ebe2a162fa8b488e6d5245bc8642103",
+   "video_sha256": "f7b57623e523cdc778cc90a882d12104b0da5772af4801c6e0ebfd62fab5c629",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/springboard_3m/trajectories/embodied-olympics-fun-springboard-3m-i00-standard--cwksgni.json b/phaseR/embodied-olympics-fun/springboard_3m/trajectories/embodied-olympics-fun-springboard-3m-i00-standard--cwksgni.json
new file mode 100644
index 0000000000000000000000000000000000000000..8498f03bd57cf1cdcafec605b9ea657ad295363f
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/springboard_3m/trajectories/embodied-olympics-fun-springboard-3m-i00-standard--cwksgni.json
@@ -0,0 +1,238 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-springboard-3m-i00-standard--cwksgni",
+ "taskKey": "embodied-olympics-fun/springboard_3m",
+ "dag_sha256": "da2d305d2df479b38c8f6f8ea05ea8ab8944e53746c08b499d4a5b65dee06ba4",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 205,
+   "sim_steps": 354,
+   "calls": 4,
+   "rows": 354,
+   "resets": 1,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 6205.7,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.093,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.329,
+   "peak_cpus": 1.7,
+   "peak_pids": 76,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-springboard-3m-i00-standard--cwksgni.mp4",
+  "duration_s": 2.08,
+  "fps": 25.0,
+  "frames": 52,
+  "sha256": "f9fc96063977d7225d8206d0b4cdc7cd0ab40148f4be8fdcf17386e17df10d6f"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "forward_straight_dive_completed": {
+    "method": "visual",
+    "intervals": [
+     [
+      1.84,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 1.84,
+      "log_step": 245,
+      "text_zh": "观察：视频 0.0-1.2 s 机器人在板端起跳并腾空前翻，约视频 1.7-1.9 s 头手朝下入水，最后一帧水面仅余小水花。agent 自述 live 服务 outcome=success、stages_reached 3/3（log step 245）；评分器字段 success=1。推断：入水时刻约 1.84 s（对照 oracle 同一机位），远景细节无法逐项核对。"
+     }
+    ]
+   }
+  },
+  "constraints": {},
+  "contact_actual": {
+   "forward_straight_dive_completed": [
+    {
+     "who": "腿/躯干",
+     "how": "支撑",
+     "note_zh": "观察：视频起始阶段双脚在跳板上蹬离起跳（远景较小，细节不清）。"
+    }
+   ]
+  }
+ },
+ "native_metrics": [
+  {
+   "name": "somersault_deg",
+   "value": null,
+   "unit": "deg",
+   "threshold": 180,
+   "direction": "target"
+  },
+  {
+   "name": "entry_distance_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": "target"
+  },
+  {
+   "name": "entry_tilt_deg",
+   "value": null,
+   "unit": "deg",
+   "threshold": 15,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_hip",
+   "value": null,
+   "unit": "deg",
+   "threshold": 160,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_knee",
+   "value": null,
+   "unit": "deg",
+   "threshold": 160,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_feet",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.12,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_toes",
+   "value": null,
+   "unit": "rad",
+   "threshold": 0.3,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_beyond",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.15,
+   "direction": "higher_better"
+  },
+  {
+   "name": "entry_apart",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.2,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_off",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.15,
+   "direction": "lower_better"
+  },
+  {
+   "name": "entry_twist",
+   "value": null,
+   "unit": "deg",
+   "threshold": 20,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "观察：graded 视频 2.08 s 内机器人从三米板起跳、前翻并头手先入水，评分器 success=1（stages 3/3）。唯一节点 前直体跳水 在约视频 1.84 s 达成，final 为真。缺口：远景下机器人很小，各项姿态指标（native_metrics 数值）run.json 未报告，故为 null。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "单节点 前直体跳水 达成、final=1 与评分器 success=1 一致；视频 1.84 s 时脚刚到水面，2.04 s 仅剩水面小残影，时间在约 1 秒内；contact_actual 腿/躯干·支撑 合理，无失败层，无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:57:01Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-springboard-3m-i00-standard--cwksgni/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-springboard-3m-i00-standard--cwksgni/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "066d4d4c89b6ff4a867b32d12e172e79f07e8aceb4145e4cca02e831334e52d7",
+   "video_sha256": "f9fc96063977d7225d8206d0b4cdc7cd0ab40148f4be8fdcf17386e17df10d6f",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/standing_long_jump/trajectories/embodied-olympics-fun-standing-long-jump-i00-privileged--6xukdgd.json b/phaseR/embodied-olympics-fun/standing_long_jump/trajectories/embodied-olympics-fun-standing-long-jump-i00-privileged--6xukdgd.json
new file mode 100644
index 0000000000000000000000000000000000000000..7dc95cf5da9f395cf4fe1a3b125e85cf824d5ef7
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/standing_long_jump/trajectories/embodied-olympics-fun-standing-long-jump-i00-privileged--6xukdgd.json
@@ -0,0 +1,148 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-standing-long-jump-i00-privileged--6xukdgd",
+ "taskKey": "embodied-olympics-fun/standing_long_jump",
+ "dag_sha256": "2d039bd85a35a257f1b909391d7b8f24129d9162d705f51bd3207cbeea156c33",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 297,
+   "sim_steps": 297,
+   "horizon": 2000,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 3,
+   "stages_total": 3,
+   "fell": 0,
+   "replay_s": 9.03,
+   "bad_trajectory": 0,
+   "metric_air_s": 0.589,
+   "metric_mark_m": 1.6,
+   "metric_rise_m": 0.437,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 67,
+   "agent_resets": 13074,
+   "agent_replays": 2,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 3.614,
+   "peak_processes": 27,
+   "peak_disk_gb": 0.832,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-standing-long-jump-i00-privileged--6xukdgd/replay.mp4",
+  "duration_s": 3.0,
+  "fps": 25.0,
+  "frames": 75,
+  "sha256": "cf69d5f47db3de3aa1b7c7f471504f961aab9b26afe39b1ddfd00b5f033ea7fc"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "standing_long_jump_completed": {
+    "method": "visual",
+    "intervals": [
+     [
+      2.9,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 2.9,
+      "log_step": null,
+      "text_zh": "观察：视频约 0.6-1.0 s 机器人双脚离地前跃，约 1.2 s 在黄线远侧屈膝落地，约 1.8 s 后直立，至 3.0 s 视频结束一直站稳、无跌倒；评分器字段 success=1、stages_reached 3/3、fell=0、metric_mark_m=1.6、air_s=0.589。推断：完成时刻为落地后连续站稳 1 s，约视频 2.9 s（297 控制步×0.01 s≈2.97 s），略不精确。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "机器人双脚蹬地起跳、以脚落地并向前调整脚步站稳；视频可见脚承重，无手部接触。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": null,
+ "summary_zh": "成功：唯一节点“立定跳远站稳”在视频约 2.9 s 达成（起跳约 0.7 s，落地约 1.2 s，越过黄线，之后站稳至视频结束），评分器 mark 1.6 m、腾空 0.589 s、未跌倒。时间点由视频与评分器步数推断，无状态日志。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "单节点成功标注与视频和评分器一致：起跳约0.7 s、越黄线落地约1.2 s、站稳至视频3.0 s；完成时刻约视频2.9 s（297步×0.01 s≈2.97 s）有推断标注；contact_actual 腿/躯干·支撑正确。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:34Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-standing-long-jump-i00-privileged--6xukdgd/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-standing-long-jump-i00-privileged--6xukdgd/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "aae0a2f0706f8db4aa5d5977db653e09115677785a33762031d6b92ed0a8303e",
+   "video_sha256": "cf69d5f47db3de3aa1b7c7f471504f961aab9b26afe39b1ddfd00b5f033ea7fc",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/standing_long_jump/trajectories/embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx.json b/phaseR/embodied-olympics-fun/standing_long_jump/trajectories/embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx.json
new file mode 100644
index 0000000000000000000000000000000000000000..fd5755734c12000dd034db1669e61c4039c054ef
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/standing_long_jump/trajectories/embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx.json
@@ -0,0 +1,259 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx",
+ "taskKey": "embodied-olympics-fun/standing_long_jump",
+ "dag_sha256": "2d039bd85a35a257f1b909391d7b8f24129d9162d705f51bd3207cbeea156c33",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 136,
+   "sim_steps": 136,
+   "calls": 9,
+   "rows": 136,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 6909.4,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.926,
+   "peak_processes": 11,
+   "peak_disk_gb": 0.354,
+   "peak_cpus": 2.33,
+   "peak_pids": 108,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx.mp4",
+  "duration_s": 1.4,
+  "fps": 25.0,
+  "frames": 35,
+  "sha256": "7a38db72005528762ea37aafca54a1f98560b44d369ad0a094c0fdeee8f5743a"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "standing_long_jump_completed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.2,
+      "log_step": null,
+      "text_zh": "观察：视频 0–约 0.7 s 机器人直立，约 0.8 s 起前倾下蹲，0.96–1.2 s 躯干大幅前倾，1.2–1.4 s 身体近水平扑向地面，手臂/躯干触地，未见双脚站稳；视频 1.4 s 结束时仍在黄线近侧。未达成。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：双脚蹬地起跳；随后躯干前倾，身体扑倒，躯干和手臂触地，未形成脚先着地的站稳。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "mark_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 1.32,
+   "direction": "higher_better"
+  },
+  {
+   "name": "air_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 0.1,
+   "direction": "higher_better"
+  },
+  {
+   "name": "rise_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "standing_long_jump_completed",
+   "t": 1.2,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 1.0–1.4 s 机器人起跳后躯干过度前倾，身体接近水平向前扑倒，未以脚先着地、未越黄线并站稳；评分器 first_reason_code=2（字段值，含义未核实）。"
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 1.3,
+    "log_step": null,
+    "evidence_zh": "观察：视频 1.2–1.4 s 躯干近水平前扑，失去支撑，视频在扑倒中结束。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "evidence_zh": "推断：agent 只开环回放预设关节轨迹（无姿态反馈）；run.json log step 196 两组对比：Clean t=0 组第 40 步真实 pt_r +0.667 vs 本地仿真 pt_s +0.403；加 60 步 Q0 保持组第 40/60 步 pt_s +0.501/+1.056，对比真实 +0.667/+1.368，差距明显缩小。开环轨迹按本地仿真调参、实机前倾更大，是仿真-实机差异/起始时序不匹配还是缺反馈纠正，无法区分，故子类不明，只归“控制”。该对比是 agent 自己的本地仿真与真实读数，未经独立状态日志核实。",
+    "source_zh": "日志",
+    "item": "其他",
+    "note_zh": "子类不明：无法区分仿真-实机差异、缺反馈或轨迹设计，故停在大类“控制”"
+   }
+  ],
+  "gap_zh": "视频仅 1.4 s，无状态日志；根因为依据 run.json 日志对比的推断，子类不明（仿真-实机差异 / 缺反馈 / 轨迹设计均未排除）；native_metrics 的数值 run.json 未报告。",
+  "unmet": {
+   "standing_long_jump_completed": false
+  }
+ },
+ "summary_zh": "观察：视频 0–1.4 s 内机器人下蹲起跳后前倾失衡，身体近水平扑倒，未脚先着地站稳，立定跳远站稳节点未达成（0/1）。断点在跳跃动作中，现象为全身·平衡；根因（推断）：控制（子类不明）——开环预设轨迹下真实躯干前倾大于本地仿真，差异含仿真-实机/起始时序不匹配，无法判定是否缺反馈（run.json log step 196）。缺口：视频极短，无状态日志。",
+ "conf": 6,
+ "review": {
+  "status": "disputed",
+  "rounds": 3,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点 standing_long_jump_completed、全身·平衡现象与视频一致；但证据里的起始时间 0.4 s 与帧不符，且 agent 日志已足以判断根因，根因层不应以“未查看日志”为由留空。",
+    "n_issues": 2,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "视频 0–1.4 s 与节点未达成、断点 standing_long_jump_completed、全身·平衡现象一致；唯一问题是根因子类“反馈控制”的推断过强：log 里同一对比的另一组（60 步 Q0 保持）显示差异大幅缩小，说明真实与本地仿真的差异部分是起始时序/模型不匹配，不能判定为缺少反馈纠正。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 3,
+    "kind": "confirm",
+    "verdict": "fix",
+    "summary_zh": "I1 的证据、summary_zh、gap_zh 已按审查意见改为“子类不明”，内容正确；但根因标签仍写成 item \"其他\"，既与回复所称“去掉 item”不符，也不符合 docs/03_trajectory.md 规定的“子类不明时停在 item 控制”。",
+    "n_issues": 1,
+    "n_checks": 1
+   }
+  ],
+  "open_zh": [
+   "[I1·第3轮未确认] evidence_zh 已补两组对比（Clean t=0 与 60 步 Q0 保持，数值与 step 196 一致），summary_zh、gap_zh 同步并仍标推断，这部分正确。但 root_causes[0] 现为 {cat:控制, item:其他, note_zh:子类不明…}，而规则要求子类不明时用 item \"控制\"（审查建议也是 \"控制\"）；回复称“去掉 item”与文件不一致，改动未按规则完整应用。"
+  ],
+  "confirm": {
+   "verdict": "not_confirmed",
+   "summary_zh": "I1 的证据、summary_zh、gap_zh 已按审查意见改为“子类不明”，内容正确；但根因标签仍写成 item \"其他\"，既与回复所称“去掉 item”不符，也不符合 docs/03_trajectory.md 规定的“子类不明时停在 item 控制”。",
+   "checks_ok": "0/1",
+   "n_factual_errors": 0
+  }
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T21:00:27Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx/04_review_r2"
+   },
+   {
+    "stage": "reconcile",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx/05_reconcile_r2"
+   },
+   {
+    "stage": "confirm",
+    "round": 3,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-standing-long-jump-i00-standard--fc3prgx/06_confirm_r3"
+   }
+  ],
+  "confirm_prompt": "prompts/06_confirm.md",
+  "confirm_prompt_sha256": "b9f56c095f7ea5e609e58bb21e076f565766c370333d71c7a3512cffb4d78267",
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "80d1b28073c2b06cd31226df21f1cc78c71d2a195e2f3a43d4bb4d7531f40976",
+   "video_sha256": "7a38db72005528762ea37aafca54a1f98560b44d369ad0a094c0fdeee8f5743a",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/steeplechase/trajectories/embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2.json b/phaseR/embodied-olympics-fun/steeplechase/trajectories/embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2.json
new file mode 100644
index 0000000000000000000000000000000000000000..e8ad0aaca17a193b852877cd1f0678a724fa4b60
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/steeplechase/trajectories/embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2.json
@@ -0,0 +1,262 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2",
+ "taskKey": "embodied-olympics-fun/steeplechase",
+ "dag_sha256": "d1654f75d2327bdc38995638268d9415b735f4a2b26d39708a7baeb8fa7d5b71",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 4168,
+   "sim_steps": 4168,
+   "calls": 54,
+   "rows": 4168,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7170.7,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 4,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.469,
+   "peak_processes": 10,
+   "peak_disk_gb": 1.522,
+   "peak_cpus": 2.98,
+   "peak_pids": 107,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2.mp4",
+  "duration_s": 41.72,
+  "fps": 25.0,
+  "frames": 1043,
+  "sha256": "ca7357a77a7b6552a830af21be3076218bfa81c2d1c2e1726713e7800cac1c52"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 2,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hurdle_cleared": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 22.3,
+      "log_step": null,
+      "text_zh": "观察：视频 17-21 s 机器人站在白色横杆前紧贴横杆，约 22 s 向后倒下，之后到视频结束（41.7 s）一直躺在栏前跑道上，未越过横杆；与 oracle 越栏后位于坑侧不同。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 0-21 s 双脚在栏前跑道行走/蹬地；约 22 s 后倒地，躯干躺在跑道上。"
+     }
+    ]
+   },
+   "pit_crossed_standing": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 41.6,
+      "log_step": null,
+      "text_zh": "观察：视频末帧机器人仍躺在栏前，未进入蓝色坑、未出坑；前置节点未达成。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "rear_x_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 13.66,
+   "direction": "higher_better"
+  },
+  {
+   "name": "time_s",
+   "value": null,
+   "unit": "s",
+   "threshold": null,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hurdle_cleared",
+   "t": 22.0,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 17-21 s 机器人贴着横杆前站立，约 22 s 向后倒下并躺在栏前至结束（41.7 s）；越栏未完成。",
+   "log_step": null
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 22.3,
+    "log_step": null,
+    "evidence_zh": "观察：视频约 22 s 机器人在栏前失去支撑向后摔倒，之后未起身。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "观察：视频约 21.8-22.6 s 机器人在栏前起跳动作后向后翻倒；log step 176-177 实机执行 gen.execute_jump_and_exit（step 173 写入的开环 lean-crouch-push-tuck-land 轨迹），step 177 的 task-346 日志显示起跳后 yaw=-115.1°、up_z=-0.011（躯干近水平），随后站不起来；step 210 本地仿真复现 up_z 翻到 -0.977（会话测试，仅作旁证）。推断：开环起跳轨迹产生后仰翻转；具体子类型不明。",
+    "source_zh": "日志 step 173-177"
+   }
+  ],
+  "gap_zh": "无状态日志；rear_x_m 等评分器数值未在 run.json 报告，native_metrics 为 null；agent 最终交付的轨迹内容仅由视频确认，倒下的具体原因（起跳失败还是撞杆）未能从帧分辨。",
+  "unmet": {
+   "hurdle_cleared": false,
+   "pit_crossed_standing": false
+  }
+ },
+ "summary_zh": "观察：机器人走到栏前（视频约 17-21 s）后向后摔倒，躺在栏前直到视频结束，越栏与出坑两个节点均未达成（0/2）。断点在越栏步骤，现象为全身·平衡；根因只能归到控制（log step 173-177 实机执行的开环起跳轨迹），子类型不明。缺少状态日志。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 3,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "WHERE 标签（越栏未达成、0/2、断点 hurdle_cleared 约 22 s、全身·平衡、contact_actual 腿/躯干·支撑）与视频一致。唯一问题：根因证据引用的是失败之后的本地 MuJoCo 会话测试（log step 289-298），而不是被评分的那次实机起跳（log step 173-177）；应改引实际执行的起跳。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "WHERE 标签与第 1 轮已核对的视频一致；仅 summary_zh 仍引用已被替换的 log step 289-298。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 3,
+    "kind": "confirm",
+    "verdict": "agree",
+    "summary_zh": "I1 已正确应用：summary_zh 现引用 log step 173-177，与 root_causes[0] 一致，未见其他改动。",
+    "n_issues": 0,
+    "n_checks": 1
+   }
+  ],
+  "open_zh": [],
+  "confirm": {
+   "verdict": "confirmed",
+   "summary_zh": "I1 已正确应用：summary_zh 现引用 log step 173-177，与 root_causes[0] 一致，未见其他改动。",
+   "checks_ok": "1/1",
+   "n_factual_errors": 0
+  }
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:40Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2/04_review_r2"
+   },
+   {
+    "stage": "reconcile",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2/05_reconcile_r2"
+   },
+   {
+    "stage": "confirm",
+    "round": 3,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-steeplechase-i00-standard--jhmlpm2/06_confirm_r3"
+   }
+  ],
+  "confirm_prompt": "prompts/06_confirm.md",
+  "confirm_prompt_sha256": "b9f56c095f7ea5e609e58bb21e076f565766c370333d71c7a3512cffb4d78267",
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ece6561271c41745c75b258a8590a4bae3154ae2e346aabac3b858fcbc98a3f4",
+   "video_sha256": "ca7357a77a7b6552a830af21be3076218bfa81c2d1c2e1726713e7800cac1c52",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/table_tennis_return/trajectories/embodied-olympics-fun-table-tennis-return-i00-privileged--umujtpn.json b/phaseR/embodied-olympics-fun/table_tennis_return/trajectories/embodied-olympics-fun-table-tennis-return-i00-privileged--umujtpn.json
new file mode 100644
index 0000000000000000000000000000000000000000..b452b1e0855b9b6a534a3cd26879a34b34ee219c
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/table_tennis_return/trajectories/embodied-olympics-fun-table-tennis-return-i00-privileged--umujtpn.json
@@ -0,0 +1,162 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-table-tennis-return-i00-privileged--umujtpn",
+ "taskKey": "embodied-olympics-fun/table_tennis_return",
+ "dag_sha256": "7c99ad814d8cc0948a54e1e64bae8599f94aff87d6cc23f6cd5219a93aa00bbb",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 479,
+   "sim_steps": 479,
+   "horizon": 800,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 3,
+   "stages_total": 3,
+   "fell": 0,
+   "replay_s": 12.72,
+   "bad_trajectory": 0,
+   "metric_grip_break_max_s": 0.006,
+   "metric_landing_x_m": 2.573,
+   "metric_landing_y_m": 0.018,
+   "metric_net_touch": 0.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 882,
+   "agent_resets": 0,
+   "agent_replays": 7,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 4.114,
+   "peak_processes": 20,
+   "peak_disk_gb": 0.901,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-table-tennis-return-i00-privileged--umujtpn/replay.mp4",
+  "duration_s": 4.8,
+  "fps": 25.0,
+  "frames": 120,
+  "sha256": "a0633cf7417fcd4516ff1485e241f61cffba20894e763241e33f27a3ee3afeeb"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "return_landed": {
+    "method": "visual",
+    "intervals": [
+     [
+      4.72,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 4.72,
+      "log_step": 49,
+      "text_zh": "观察：视频 4.3 s 前后右手所持红拍向前上挥击，小球越过白网飞向发球机一侧；视频约 4.7-4.8 s 小球落在对方蓝色台面。评分器字段 success=1、stages 3/3、落点 x=2.573 m、net_touch=0，持拍中断最大 0.006 s。推断：落台时间约 4.7 s，精度约 0.2 s（球很小）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "其他",
+      "note_zh": "候选：击打；右手握持的球拍拍面在视频约 4.3 s 击打来球，球拍全程未离手。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "双脚支撑身体，未摔倒（fell=0）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "grip_break_max_s",
+   "value": 0.006,
+   "unit": "s",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：右手持拍在视频约 4.3 s 击球，球过网落对方台面（评分器 stages 3/3，落点 x=2.573 m，无触网，持拍中断最大 0.006 s）。落台精确时刻因球小只能估计为视频约 4.7 s。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "成功运行，单节点 return_landed 终值 true，落台约 4.7 s 与帧一致，评分器 stages 3/3；接触标注（击打=手持工具·其他、腿/躯干支撑）合理，无失败层。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:58Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-table-tennis-return-i00-privileged--umujtpn/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-table-tennis-return-i00-privileged--umujtpn/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9fc92ca6eef19737253b3c3df6d89d64a257cdecd4940b94ad8d5a41764153fc",
+   "video_sha256": "a0633cf7417fcd4516ff1485e241f61cffba20894e763241e33f27a3ee3afeeb",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/table_tennis_return/trajectories/embodied-olympics-fun-table-tennis-return-i00-realtime--v6mzsq9.json b/phaseR/embodied-olympics-fun/table_tennis_return/trajectories/embodied-olympics-fun-table-tennis-return-i00-realtime--v6mzsq9.json
new file mode 100644
index 0000000000000000000000000000000000000000..68277869e393379f668cda0b0299ba13864e2de4
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/table_tennis_return/trajectories/embodied-olympics-fun-table-tennis-return-i00-realtime--v6mzsq9.json
@@ -0,0 +1,221 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-table-tennis-return-i00-realtime--v6mzsq9",
+ "taskKey": "embodied-olympics-fun/table_tennis_return",
+ "dag_sha256": "7c99ad814d8cc0948a54e1e64bae8599f94aff87d6cc23f6cd5219a93aa00bbb",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 439,
+   "sim_steps": 1790,
+   "calls": 152,
+   "rows": 18160,
+   "resets": 4,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 6908.5,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "realtime": 1,
+   "slowdown": 1.0,
+   "clock_s": 28.871,
+   "max_lag_s": 0.5102,
+   "stalls": 19,
+   "stalled_s": 9.575,
+   "stalled_share": 0.3316,
+   "held_steps": 439,
+   "late_steps": 0,
+   "requests_refused": 1,
+   "request_busy_s": 3.362,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.982,
+   "peak_processes": 8,
+   "peak_disk_gb": 0.428,
+   "peak_cpus": 1.39,
+   "peak_pids": 175,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-table-tennis-return-i00-realtime--v6mzsq9.mp4",
+  "duration_s": 4.4,
+  "fps": 25.0,
+  "frames": 110,
+  "sha256": "a630e570b2416e643ce338e3a16a672dac89d1d88603c81d172a41e64d0b30b1"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "return_landed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 4.3,
+      "log_step": null,
+      "text_zh": "观察：视频 0-2.2 s 机器人右手持红色球拍摆臂；视频 ≈2.3-2.4 s 球拍离开右手，落在手旁台面上并留在那里至 4.4 s；其间未见球拍击球；2.5-4.4 s 小球仅见于远端发球机一侧，未见球越网落对台。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手指/夹爪",
+      "how": "抓握",
+      "note_zh": "观察：右手 Dex3 手指握持球拍摆动，视频 ≈2.3 s 球拍脱手；未见球拍或手与球接触。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "双脚着地支撑，全程站立。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "grip_break_max_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "return_landed",
+   "t": 2.3,
+   "log_step": null,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 ≈2.3 s 球拍脱离右手落在台面，之后直到视频 4.4 s 结束都没有击球和回球落对台；grader 字段 first_reason_code=2、success=0。"
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 2.3,
+    "log_step": null,
+    "evidence_zh": "观察：视频 1.6-2.2 s 右手持拍，≈2.3-2.4 s 球拍离手，落在手旁台面并留在那里，未见球拍击球。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "无状态日志，视频分辨率低（640x360），球极小，击球时刻与脱手原因（握力不足或挥拍加速度）无法确认；agent 日志只有大量离线仿真测试，无法对应图形中的脱手，故根因留空。",
+  "unmet": {
+   "return_landed": false
+  }
+ },
+ "summary_zh": "观察：回球落对台未达成（final false）。机器人右手持拍摆动，视频 ≈2.3 s 球拍脱手，球拍落在手旁台面，之后空手直至视频 4.4 s，未见击球与回球。断点为回球落对台的击球操作中，现象为球拍没保持住；因无状态日志且球与握持细节看不清，根因未判。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点 return_landed 操作中、现象 接触·没保持住（球拍≈2.3 s 脱手）与帧一致；仅证据描述有误：球拍脱手后并非『画面无球拍』，而是落在手旁台面上。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点 return_landed 操作中、现象 接触·没保持住（视频≈2.3 s 球拍脱手落台面）与帧一致；根因留空有 gap_zh，合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:14Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-table-tennis-return-i00-realtime--v6mzsq9/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-table-tennis-return-i00-realtime--v6mzsq9/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-table-tennis-return-i00-realtime--v6mzsq9/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-table-tennis-return-i00-realtime--v6mzsq9/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "a427b21b287491cd3d9e56f154949ee17dc1d8508cdc4fc4d99137a28eb8638c",
+   "video_sha256": "a630e570b2416e643ce338e3a16a672dac89d1d88603c81d172a41e64d0b30b1",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/taekwondo_kick/trajectories/embodied-olympics-fun-taekwondo-kick-i00-privileged--sqd8zus.json b/phaseR/embodied-olympics-fun/taekwondo_kick/trajectories/embodied-olympics-fun-taekwondo-kick-i00-privileged--sqd8zus.json
new file mode 100644
index 0000000000000000000000000000000000000000..dfd07000eefa2b2b79ce9ae969abec2d2ef7a88c
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/taekwondo_kick/trajectories/embodied-olympics-fun-taekwondo-kick-i00-privileged--sqd8zus.json
@@ -0,0 +1,168 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-taekwondo-kick-i00-privileged--sqd8zus",
+ "taskKey": "embodied-olympics-fun/taekwondo_kick",
+ "dag_sha256": "63ae944f698e6469950e0e47b5b44e99a40b4ebb86df1180bd7baa564cd727b9",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 618,
+   "sim_steps": 618,
+   "horizon": 1500,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 2,
+   "stages_total": 2,
+   "fell": 0,
+   "replay_s": 6.72,
+   "bad_trajectory": 0,
+   "metric_contact_s": 0.034,
+   "metric_impulse_n_s": 13.27,
+   "metric_peak_force_n": 966.0,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 46,
+   "agent_resets": 1307,
+   "agent_replays": 9,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 2.561,
+   "peak_processes": 19,
+   "peak_disk_gb": 0.992,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "session_log_tampered": 0,
+   "session_log_from_monitor": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-taekwondo-kick-i00-privileged--sqd8zus/replay.mp4",
+  "duration_s": 6.2,
+  "fps": 25.0,
+  "frames": 155,
+  "sha256": "006f49df11839bd002d4f4210f987e5d5d1870f0e95599ef33da5b0d28ea4efc"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "protector_kicked_and_stood": {
+    "method": "visual",
+    "intervals": [
+     [
+      6.18,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 6.18,
+      "log_step": 76,
+      "text_zh": "观察：视频 1.6–3.6 s 机器人抬起左腿、折膝并前伸向假人护具方向（场边视角下踢击瞬间被侧向遮挡，接触帧无法逐帧确认）；视频 ≈4.0–6.2 s 机器人双脚站在红垫、躯干直立、离开假人，无跌倒。评分器字段：success=1，stages_reached 2/2（kick、stand），metric_impulse_n_s 13.27 ≥ 7.2，contact_s 0.034，fell 0。推断：连续站稳 2.0 s 在视频末尾 ≈6.18 s 完成（agent 自述完成于 6.18 s，log step 76；与视频时长 6.2 s 吻合）。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "左脚以踝下足部短促击打护具（候选名“击打”），评分器记录冲量 13.27 N·s、接触 0.034 s"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "另一只（右）脚在踢击时支撑红色垫面"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "impulse_n_s",
+   "value": 13.27,
+   "unit": "N s",
+   "threshold": 7.2,
+   "direction": "higher_better"
+  },
+  {
+   "name": "contact_s",
+   "value": 0.034,
+   "unit": "s",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：机器人（脚本化全身控制轨迹）抬左腿踢中护具（评分器冲量 13.27 N·s ≥ 7.2，接触 0.034 s），收腿后双脚站稳，视频约 6.18 s 完成，未摔倒。证据缺口：场边视角下踢击接触瞬间被遮挡，接触时刻主要来自评分器字段，视频只能确认踢前动作与随后的站稳。",
+ "conf": 7,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "标签与视频、评分器字段一致：踢击发生在视频约1.5–3.5 s，随后站稳；618个控制步×10 ms=6.18 s 与视频时长6.2 s吻合，节点区间[6.18,末尾]、contact_actual（其他·击打 + 支撑）正确，成功无失败层。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:57:08Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-taekwondo-kick-i00-privileged--sqd8zus/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-taekwondo-kick-i00-privileged--sqd8zus/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "ef5d35370f56d1abb2c8883e7f5a5a26f6b959e4134b6d319110c5e9ff772488",
+   "video_sha256": "006f49df11839bd002d4f4210f987e5d5d1870f0e95599ef33da5b0d28ea4efc",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/taekwondo_kick/trajectories/embodied-olympics-fun-taekwondo-kick-i00-standard--li5juik.json b/phaseR/embodied-olympics-fun/taekwondo_kick/trajectories/embodied-olympics-fun-taekwondo-kick-i00-standard--li5juik.json
new file mode 100644
index 0000000000000000000000000000000000000000..bdab76765cd08881be45e1a018b04b0d28d1871c
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/taekwondo_kick/trajectories/embodied-olympics-fun-taekwondo-kick-i00-standard--li5juik.json
@@ -0,0 +1,168 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-taekwondo-kick-i00-standard--li5juik",
+ "taskKey": "embodied-olympics-fun/taekwondo_kick",
+ "dag_sha256": "63ae944f698e6469950e0e47b5b44e99a40b4ebb86df1180bd7baa564cd727b9",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "live_success": 1,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 444,
+   "sim_steps": 523,
+   "calls": 229,
+   "rows": 523,
+   "resets": 1,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 5033.8,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.161,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.321,
+   "peak_cpus": 1.72,
+   "peak_pids": 78,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-taekwondo-kick-i00-standard--li5juik.mp4",
+  "duration_s": 4.48,
+  "fps": 25.0,
+  "frames": 112,
+  "sha256": "38aac6eacca7623ad01a28662e2a1118635aa86a90069c52e54d3f2d84bd7767"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "protector_kicked_and_stood": {
+    "method": "visual",
+    "intervals": [
+     [
+      4.44,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 0.86,
+      "log_step": null,
+      "text_zh": "观察：视频 0.8-0.9 s，机器人右腿屈膝抬起，脚踢到假人胸前蓝色护具，随后收腿（逐帧 25 fps 检查）。"
+     },
+     {
+      "t": 4.44,
+      "log_step": null,
+      "text_zh": "观察：视频 1.5-4.48 s 机器人双脚落在红垫上站立，离开假人，姿态直立。推断：200 个控制步（2.0 s）站稳窗口在约视频 4.44 s 完成，与评分器 success=1、stages_reached 2/2、4.44 s / 444 步（agent 最后报告引用服务读数）一致。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "其他",
+      "note_zh": "右脚（踝下足部）击打假人蓝色护具，候选名称“击打”，视频约 0.85 s。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "左脚支撑红色垫面，落地后双脚支撑站稳。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": null,
+ "summary_zh": "观察：成功，唯一节点“踢中护具并站稳”达成。视频约 0.85 s 右脚踢中护具，随后收腿落地双脚站稳，约视频 4.44 s 完成 2.0 s 站稳窗口。冲量等数值未在视频中核实，依据评分器成功判定。",
+ "conf": 8,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 8,
+    "summary_zh": "成功运行，单节点达成；踢击约0.5-0.9 s、站稳约4.44 s（控制步444=4.44 s，与视频时长4.48 s吻合），接触标注符合S16，无需修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:56:52Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-taekwondo-kick-i00-standard--li5juik/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-taekwondo-kick-i00-standard--li5juik/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "2a939f7e0e46df85545d2fc3c6e60e237c7718631ab732614ba3ce3c10631df1",
+   "video_sha256": "38aac6eacca7623ad01a28662e2a1118635aa86a90069c52e54d3f2d84bd7767",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/tennis_return/trajectories/embodied-olympics-fun-tennis-return-i00-privileged--x8vtmax.json b/phaseR/embodied-olympics-fun/tennis_return/trajectories/embodied-olympics-fun-tennis-return-i00-privileged--x8vtmax.json
new file mode 100644
index 0000000000000000000000000000000000000000..75bad7173a5a9aac28605974bc4c259adc8bef37
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/tennis_return/trajectories/embodied-olympics-fun-tennis-return-i00-privileged--x8vtmax.json
@@ -0,0 +1,184 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-tennis-return-i00-privileged--x8vtmax",
+ "taskKey": "embodied-olympics-fun/tennis_return",
+ "dag_sha256": "ccffc71d0aa2a0e2f5f39987360ec83305903d4551c5e4b8aa3a2ecae36782d8",
+ "outcome": {
+  "success": 1,
+  "reward": 1.0,
+  "source": "catalog",
+  "raw": {
+   "success": 1,
+   "replay_success": 1,
+   "deterministic": 1,
+   "n_actions": 553,
+   "sim_steps": 553,
+   "horizon": 1000,
+   "episode_over": 1,
+   "progress": 1.0,
+   "stages_reached": 4,
+   "stages_total": 4,
+   "fell": 0,
+   "replay_s": 9.07,
+   "bad_trajectory": 0,
+   "metric_grip_break_max_s": 0.001,
+   "metric_landing_x_m": 18.332,
+   "metric_landing_y_m": 2.749,
+   "grader_error": 0,
+   "video_rendered": 1,
+   "agent_sessions": 64,
+   "agent_resets": 3586,
+   "agent_replays": 22,
+   "agent_cost_self_reported": 1,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 1,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.0,
+   "peak_processes": 0.0,
+   "peak_disk_gb": 0.0,
+   "resource_emergency_kills": 0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "hf",
+  "url": "https://huggingface.co/datasets/MLL-Lab/embodied-exam-media/resolve/main/runs/embodied-olympics-fun-tennis-return-i00-privileged--x8vtmax/replay.mp4",
+  "duration_s": 5.56,
+  "fps": 25.0,
+  "frames": 139,
+  "sha256": "f163b5fd89d90f9d2bffe7a4e46e2f262043655b9e5d1937671e16c57813de9c"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 1,
+  "final": 1,
+  "nodes": {
+   "return_landed_in": {
+    "method": "visual",
+    "intervals": [
+     [
+      5.5,
+      null
+     ]
+    ],
+    "final": true,
+    "evidence": [
+     {
+      "t": 4.2,
+      "log_step": null,
+      "text_zh": "观察：视频约 4.0-4.3 s 机器人右手握持的球拍由下垂抬起上摆（击球动作），球仅为数个像素，难以逐帧追踪。"
+     },
+     {
+      "t": 5.5,
+      "log_step": null,
+      "text_zh": "推断：评分器判定成功（stages 4/4，fell 0，落点 x=18.33 m、y=2.75 m 在合法区内，grip_break_max 0.001 s）；成功后回合立即结束，故落地时刻约等于视频结束（≈5.5 s），该时刻无法在帧中直接核对。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "其他",
+      "note_zh": "右手持拍，拍面击打网球（候选：击打），视频约 4.2 s 挥拍。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "双腿支撑站立，机器人先向左转并走位（log step 15 自述），底盘无。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "grip_break_max_s",
+   "value": 0.001,
+   "unit": "s",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  },
+  {
+   "name": "landing_x_m",
+   "value": 18.332,
+   "unit": "m",
+   "threshold": [
+    12.385,
+    24.27
+   ],
+   "direction": "target"
+  },
+  {
+   "name": "landing_y_m",
+   "value": 2.749,
+   "unit": "m",
+   "threshold": [
+    -1.365,
+    6.865
+   ],
+   "direction": "target"
+  }
+ ],
+ "failure": null,
+ "summary_zh": "成功：评分器 4/4 阶段，回球落入对场（x=18.33 m，y=2.75 m）。视频中可见机器人持拍上挥（约 4.2 s），球太小，落地时刻按成功即终止推断≈视频结束（5.5 s），为主要证据缺口。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "标签与视频、评分器结果一致：成功，唯一目标节点 return_landed_in 为真，时间 5.5 s 明确标为推断，接触标签合理（持拍击打=其他，腿/躯干支撑）。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:57:00Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-tennis-return-i00-privileged--x8vtmax/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-tennis-return-i00-privileged--x8vtmax/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "a015c0b3ce2879af6091b46e765367671e05192e93ca85f9c987ff3b17d3d864",
+   "video_sha256": "f163b5fd89d90f9d2bffe7a4e46e2f262043655b9e5d1937671e16c57813de9c",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/tennis_return/trajectories/embodied-olympics-fun-tennis-return-i00-realtime--8japwdw.json b/phaseR/embodied-olympics-fun/tennis_return/trajectories/embodied-olympics-fun-tennis-return-i00-realtime--8japwdw.json
new file mode 100644
index 0000000000000000000000000000000000000000..286105db8e5c535bfb34cf7e2dd28fa27d204b39
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/tennis_return/trajectories/embodied-olympics-fun-tennis-return-i00-realtime--8japwdw.json
@@ -0,0 +1,243 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-tennis-return-i00-realtime--8japwdw",
+ "taskKey": "embodied-olympics-fun/tennis_return",
+ "dag_sha256": "ccffc71d0aa2a0e2f5f39987360ec83305903d4551c5e4b8aa3a2ecae36782d8",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 673,
+   "sim_steps": 1582,
+   "calls": 922,
+   "rows": 12001,
+   "resets": 2,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7060.5,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "realtime": 1,
+   "slowdown": 1.0,
+   "clock_s": 30.145,
+   "max_lag_s": 0.5112,
+   "stalls": 28,
+   "stalled_s": 14.112,
+   "stalled_share": 0.4681,
+   "held_steps": 249,
+   "late_steps": 2,
+   "requests_refused": 0,
+   "request_busy_s": 4.664,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.699,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.416,
+   "peak_cpus": 4.01,
+   "peak_pids": 178,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-tennis-return-i00-realtime--8japwdw.mp4",
+  "duration_s": 6.76,
+  "fps": 25.0,
+  "frames": 169,
+  "sha256": "8fdac049a3179566543b2a8ba5f4cb0cf35d4dc503d87f04e4a7a90931558214"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "return_landed_in": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 4.1,
+      "log_step": 195,
+      "text_zh": "观察：视频约 4.0 s 机器人右手持拍挥拍击球，随后球向高处飞起（约 4.3 s 在网端上方后出画面），之后在 6.76 s 视频结束前未见球越网落入远侧场地。观察：log step 195 agent 自述第3次尝试 stages_reached=2/4、球飞至 t=6.73 s，与视频长度 6.76 s 吻合；评分字段 first_reason_code=2、success=0。推断：已完成反弹与击球两阶段，未完成越网与落点。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "手持工具",
+      "how": "其他",
+      "note_zh": "观察：右手灵巧手握持球拍，拍面击打来球（候选：击打），视频约 4.0 s。"
+     },
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：双脚站立支撑挥拍，未摔倒。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "grip_break_max_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 0.05,
+   "direction": "lower_better"
+  },
+  {
+   "name": "landing_x_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": null
+  },
+  {
+   "name": "landing_y_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "return_landed_in",
+   "t": 4.5,
+   "where": "操作中",
+   "evidence_zh": "观察：约 4.0 s 击球后球快速朝网方向飞高，约 4.3 s 在网端上方、随后出画面，6.76 s 前未见其越网落入远侧场地；评分字段 stages_reached=2/4、first_reason_code=2、success=0，节点未达成。推断（仅 agent 自述 log step 196）：球高吊、滞空约 2.6 s、落点 x=0.44 m/y=6.75 m，视频未能核实。",
+   "log_step": 196
+  },
+  "phenomena": [
+   {
+    "cat": "运动",
+    "item": "位姿不准",
+    "t": 4.5,
+    "log_step": 196,
+    "evidence_zh": "观察：击球后球快速飞高并出画面，6.76 s 结束前未落入远侧场地。推断（依据 agent 自述 log step 196，视频无法核实）：回球方向/仰角错误（高吊）。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断（把握低）：agent 自述（log step 196）挥拍法线 ns_ik[2]=0.33-0.38 上扬过大；非可独立核对的服务读数，视频看不出拍面法线。挥拍已完成且握拍未丢失（log step 195）。",
+    "source_zh": "agent 自述",
+    "alt_feasible": "未知"
+   }
+  ],
+  "gap_zh": "球很小，视频中看不清球落点；落点和 grip_break 数值评分器未给出（native_metrics 为 null）。phenomenon 的 item 名按分类表近似选取，未逐项核对。 球路（高吊、落点、挥拍法线）无法由视频核实，仅为 agent 自述推断。",
+  "unmet": {
+   "return_landed_in": false
+  }
+ },
+ "summary_zh": "观察：回球任务 1 个节点，机器人完成反弹和击球（评分 stages 2/4），但回球被打成高吊，未越网落入对场，节点未达成。断点在回球落入对场的击球操作中；推断为挥拍仰角控制偏差（依据 agent 自述 log step 196）。缺口：视频中球过小，落点未直接看到。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点在 return_landed_in、击球时刻约 4.0 s 与视频相符，WHERE 标签基本正确；问题在于把视频里没有的“近乎竖直高吊”写成观察，且根因只靠 agent 自述却未标为推断。",
+    "n_issues": 2,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "第1轮问题已修正；节点未达成、断点在击球操作、观察/推断分列均与视频（击球约4.0 s，球出画面，6.76 s前无落点）一致。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:19Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-tennis-return-i00-realtime--8japwdw/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-tennis-return-i00-realtime--8japwdw/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-tennis-return-i00-realtime--8japwdw/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-tennis-return-i00-realtime--8japwdw/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "fd9cc77f0476c7382dfee726c4fd94312d5c075568b25754d40b4f441cfd2e80",
+   "video_sha256": "8fdac049a3179566543b2a8ba5f4cb0cf35d4dc503d87f04e4a7a90931558214",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/tennis_serve/trajectories/embodied-olympics-fun-tennis-serve-i00-standard--jzopm9x.json b/phaseR/embodied-olympics-fun/tennis_serve/trajectories/embodied-olympics-fun-tennis-serve-i00-standard--jzopm9x.json
new file mode 100644
index 0000000000000000000000000000000000000000..009ac1ed84bed50d07cffd3845ca19fc0df4be2d
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/tennis_serve/trajectories/embodied-olympics-fun-tennis-serve-i00-standard--jzopm9x.json
@@ -0,0 +1,189 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-tennis-serve-i00-standard--jzopm9x",
+ "taskKey": "embodied-olympics-fun/tennis_serve",
+ "dag_sha256": "75f52b1653f7c3603e4ad1e686c78bfabba387976b8d85a1aaa2f5742b889fd4",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 175,
+   "sim_steps": 175,
+   "calls": 4,
+   "rows": 175,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 5010.0,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 1.445,
+   "peak_processes": 10,
+   "peak_disk_gb": 0.363,
+   "peak_cpus": 4.02,
+   "peak_pids": 147,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-tennis-serve-i00-standard--jzopm9x.mp4",
+  "duration_s": 1.76,
+  "fps": 25.0,
+  "frames": 44,
+  "sha256": "39b7f30762458714e3e0032eed47b3070a5abaadae1e62dfc73c62e6fb546810"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "serve_in_box": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 1.7,
+      "log_step": 191,
+      "text_zh": "观察：视频 0-1.5 s 机器人左臂举到头侧、右手持拍在身侧；约视频 1.57 s 起球从左手附近放出（小黄点），视频 1.6-1.76 s 球拍从右手甩脱并向身体右侧飞离，球拍未击到球，球未过网、未入发球区。观察：log step 191 服务状态 control_steps 175（仿真 1.75 s ≈ 视频结束），done=True，reason 'failed'，outcome 'the racket left your right hand'，stages_reached 1/4。"
+     }
+    ]
+   }
+  },
+  "constraints": {},
+  "contact_actual": {
+   "serve_in_box": [
+    {
+     "who": "手持工具",
+     "how": "其他",
+     "note_zh": "观察：右手持拍摆动，视频 1.6 s 后球拍脱手，未见拍与球接触（击打未发生）。"
+    },
+    {
+     "who": "腿/躯干",
+     "how": "支撑",
+     "note_zh": "观察：双脚始终站立在底线后，视频中无摔倒。"
+    }
+   ]
+  }
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "serve_in_box",
+   "where": "操作中",
+   "t": 1.65,
+   "evidence_zh": "观察：视频约 1.6-1.76 s 右手持拍挥出时球拍脱手飞离，球在左手放出后无拍击球；log step 191 服务状态 reason=failed，outcome=“the racket left your right hand”。推断：击球步骤中途因拍脱手终止。",
+   "log_step": 191
+  },
+  "phenomena": [
+   {
+    "cat": "接触",
+    "item": "没保持住",
+    "t": 1.7,
+    "log_step": 191,
+    "evidence_zh": "观察：视频 0-1.5 s 右手握住球拍，约 1.6 s 挥拍时球拍离开右手；服务状态 outcome 为 the racket left your right hand（log step 191）。"
+   }
+  ],
+  "root_causes": [
+   {
+    "cat": "控制",
+    "item": "控制",
+    "evidence_zh": "推断：挥拍加速时握力/接触不足以保持球拍；agent 本地模型用的是与评分环境不同的手部模型（log step 258-262 在校准数字孪生），未验证握持。具体子类型不清，故只标控制。",
+    "source_zh": "日志"
+   }
+  ],
+  "gap_zh": "视频分辨率低，球拍脱手的具体原因（握力不足或挥拍过猛）及球的精确离手时刻无法精确判断；无状态日志。",
+  "unmet": {
+   "serve_in_box": false
+  }
+ },
+ "summary_zh": "观察：视频 1.76 s，机器人抛出球并挥拍，但在视频约 1.6-1.76 s 球拍从右手脱落，未击球，发球入区未达成（0/1）。断点在发球击打步骤，现象为球拍没保持住，根因只能定为控制（子类型不明）。缺口：无状态日志，视频低清。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 1,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "节点未达成、断点（serve_in_box，操作中）、现象（接触·没保持住）与视频一致：球于视频约1.56-1.6 s放出，球拍在约1.68-1.76 s从右手脱落，服务状态 control_steps 175 / 'the racket left your right hand'。根因标控制且注明子类型不清，属诚实的推断，不提出修改。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:57:24Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-tennis-serve-i00-standard--jzopm9x/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-tennis-serve-i00-standard--jzopm9x/04_review_r1"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "f52d1b2418251323354d6126faa4edb051fa406f01b4267611ddbf9447441963",
+   "video_sha256": "39b7f30762458714e3e0032eed47b3070a5abaadae1e62dfc73c62e6fb546810",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/triple_jump/trajectories/embodied-olympics-fun-triple-jump-i00-standard--f4mm5wr.json b/phaseR/embodied-olympics-fun/triple_jump/trajectories/embodied-olympics-fun-triple-jump-i00-standard--f4mm5wr.json
new file mode 100644
index 0000000000000000000000000000000000000000..f49dcb76f779295a1715e6f6fdb5f2a4c24a9b3a
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/triple_jump/trajectories/embodied-olympics-fun-triple-jump-i00-standard--f4mm5wr.json
@@ -0,0 +1,266 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-triple-jump-i00-standard--f4mm5wr",
+ "taskKey": "embodied-olympics-fun/triple_jump",
+ "dag_sha256": "0baf08d04fbf745076feddee38258dc6f6ef2b10629220cbb9e20880d00d44ba",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 709,
+   "sim_steps": 3401,
+   "calls": 38,
+   "rows": 3401,
+   "resets": 4,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7176.3,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "resource_enforced": 0,
+   "limits_violation_observed": 1,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 0.89,
+   "peak_processes": 13,
+   "peak_disk_gb": 2.362,
+   "peak_cpus": 4.21,
+   "peak_pids": 107,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0,
+   "over_cpus": 1
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-triple-jump-i00-standard--f4mm5wr.mp4",
+  "duration_s": 7.12,
+  "fps": 25.0,
+  "frames": 178,
+  "sha256": "e529c32dc3151ca850081d1a0c32389f943229f24aaa43835e0f4a3823ad4b01"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 3,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "hop_landed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 7.0,
+      "log_step": 212,
+      "text_zh": "观察：视频 0-5.3 s 只有空跑道；约 5.4 s 机器人从左侧入画沿跑道奔跑，6.8 s 后脚离地、身体前倾腾空于起跳线附近，视频在 7.12 s 结束，未见落地（非“倒下”）。hop_landed 保持未达成：视频内未见在线后着地，末次 episode 结果未见。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：视频 5.4-6.8 s 双脚交替支撑奔跑，6.8 s 后脚离地腾空，视频在空中结束（7.12 s），未见落地。"
+     }
+    ]
+   },
+   "step_landed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 7.0,
+      "log_step": 212,
+      "text_zh": "观察：视频结束时机器人仍在起跳线附近空中，未见第一跳落地，故未进入跨步跳。"
+     }
+    ]
+   },
+   "jump_in_sand": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 7.0,
+      "log_step": 212,
+      "text_zh": "观察：视频结束时机器人仍在起跳线附近，从未到达沙坑（沙坑始终空）。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "hop_air_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 0.05,
+   "direction": "higher_better"
+  },
+  {
+   "name": "step_air_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 0.05,
+   "direction": "higher_better"
+  },
+  {
+   "name": "jump_air_s",
+   "value": null,
+   "unit": "s",
+   "threshold": 0.05,
+   "direction": "higher_better"
+  },
+  {
+   "name": "hop_rise_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.05,
+   "direction": "higher_better"
+  },
+  {
+   "name": "hop_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": "higher_better"
+  },
+  {
+   "name": "step_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": "higher_better"
+  },
+  {
+   "name": "mark_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 7.0,
+   "direction": "higher_better"
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "hop_landed",
+   "t": 6.9,
+   "log_step": 212,
+   "where": "操作中",
+   "evidence_zh": "观察：视频约 5.4 s 起机器人沿跑道奔跑，6.8 s 后越过起跳线附近离地腾空，视频在空中结束，未见落地。此前 episode 2、3 的服务结果为“a foot on or over the take-off line”（log step 203、206，属 agent 会话，仅作佐证）；本视频对应末次 episode，log 无其结果。run 被时限截断（wall_s 7201 ≥ 7200 s，step 212 于 7160 s 才启动脚本）。"
+  },
+  "phenomena": [],
+  "root_causes": [],
+  "gap_zh": "视频在腾空中途结束（7.12 s），末次 episode 的结果（是否合格起跳、是否落地）既不在 log 也不在视频中；run 在时限处被截断（wall_s 7201 / 限 7200 s，Episode 5 于 7160 s 才启动），属外因截断；故不标注现象与根因，不把 agent 会话里 step 203/206 的失败当作本次观测。无状态日志；机器人较小，足部接触细节不可见。",
+  "unmet": {
+   "hop_landed": false,
+   "step_landed": false,
+   "jump_in_sand": false
+  }
+ },
+ "summary_zh": "观察：视频 7.12 s，前 5 s 为空跑道，机器人约 5.4 s 入画奔跑，6.8 s 后在起跳线附近离地腾空，视频在空中结束，未见落地；三个节点均未达成（0/3）。断点在 hop_landed，但 run 被时限截断（7201/7200 s），末次 episode 结果未见，现象与根因无证据，留空（见 gap_zh）。",
+ "conf": 3,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "视频 6.7-7.12 s 逐帧（25 fps 裁切放大）显示机器人越过起跳线时已离地腾空（影子与脚分离、身体前倾上升），视频在空中结束，并没有“在线处倒下、躯干触地”；因此“平衡”现象与“控制”根因缺乏证据，且视频末尾未见落地，hop_landed 无法由视频判定。run.json 显示 wall_s 7201 ≥ 限时 7200 s，Episode 5 在 7160 s 才启动脚本，log 无结果，属被时限截断；",
+    "n_issues": 4,
+    "n_major": 2
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 6,
+    "summary_zh": "标签与视频一致：视频 5.4 s 起奔跑，约 6.8 s 后在起跳线附近腾空，7.12 s 结束仍在空中，三节点未达成；run 在 wall_s 7201/7200 s 被时限截断，末次 episode 无结果，现象与根因留空并有 gap_zh，合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T21:09:54Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-triple-jump-i00-standard--f4mm5wr/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-triple-jump-i00-standard--f4mm5wr/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-triple-jump-i00-standard--f4mm5wr/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-triple-jump-i00-standard--f4mm5wr/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "07ef9ecb006f95d625d2f5e9bb362afb4c1cc7c15c9579eefbc73b746236b1f5",
+   "video_sha256": "e529c32dc3151ca850081d1a0c32389f943229f24aaa43835e0f4a3823ad4b01",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/vault/trajectories/embodied-olympics-fun-vault-i00-standard--vzeatet.json b/phaseR/embodied-olympics-fun/vault/trajectories/embodied-olympics-fun-vault-i00-standard--vzeatet.json
new file mode 100644
index 0000000000000000000000000000000000000000..10a7d54852ae65dddf20d51d2a976330590a4eec
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/vault/trajectories/embodied-olympics-fun-vault-i00-standard--vzeatet.json
@@ -0,0 +1,197 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-vault-i00-standard--vzeatet",
+ "taskKey": "embodied-olympics-fun/vault",
+ "dag_sha256": "a1f52ddbddb968aa861d1edbedb34b470d89b71f3c09eecdf567d803c9dea268",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 1675,
+   "sim_steps": 4575,
+   "calls": 26,
+   "rows": 12,
+   "resets": 1,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 7177.7,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 3,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0,
+   "peak_memory_gb": 1.546,
+   "peak_processes": 8,
+   "peak_disk_gb": 0.491,
+   "peak_cpus": 1.85,
+   "peak_pids": 75,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-vault-i00-standard--vzeatet.mp4",
+  "duration_s": 16.76,
+  "fps": 25.0,
+  "frames": 419,
+  "sha256": "6ae1fbc151b6859ddcf607dbe8dc17586df0af76bef32163de2a8b506bf70273"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "vault_landed_stable": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 16.7,
+      "log_step": 250,
+      "text_zh": "观察：视频 0 s 机器人站在跑道最左端，到视频 16.7 s 仍在跑道左段直立行走，离跳板和跳马台还很远（约四分之一路程），全程未起跳、未触台面、未越过、未落垫。推断：撑越跳马未达成，评分器 success=0。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "腿/躯干",
+      "how": "支撑",
+      "note_zh": "观察：机器人双脚在地面上行走助跑，全程未接触跳马台和落地垫。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [],
+ "failure": {
+  "breakpoint": {
+   "ref": "vault_landed_stable",
+   "t": 16.7,
+   "log_step": 250,
+   "where": "操作中",
+   "evidence_zh": "观察：视频 0-16.76 s 机器人只沿跑道向跳板步行，结束时仍离跳板很远，起跳、撑台、越过、落垫都没开始。agent 最后一步（log step 250）在实机上运行 gen.py 生成的轨迹，未见其结果。"
+  },
+  "unmet": {
+   "vault_landed_stable": false
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "步态",
+    "t": 16.7,
+    "log_step": 250,
+    "evidence_zh": "观察：视频 0-16.7 s 机器人直立步行，速度很慢，16.76 s 视频结束时只走完约四分之一的助跑距离，没有摔倒，也没到跳板。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "根因证据不足：交付的 gen.py 内容与步行指令未在日志中完整显示（log step 250 无输出），无法判断步行为何这么慢（速度指令、控制或脚本问题）；agent 在 log step 242 的 Gemini 过滤器拦截（log step 243）后仍继续并交付，不属外部因素截断。无状态日志，节点依据视频。"
+ },
+ "summary_zh": "观察：graded 视频 16.76 s 内机器人只从跑道起点慢速步行了约四分之一路程，未到跳板，撑越跳马的唯一节点未达成（success=0）。断点在撑越跳马的助跑阶段，现象为步态过慢；根因缺证据。agent 的大量仿真调试（log step 185-248）并未体现在这段视频中。",
+ "conf": 6,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 7,
+    "summary_zh": "节点未达成、final=false、步态现象与视频一致（机器人 0-16.7 s 慢速步行，仅走约四分之一路程）。唯一问题：断点 where 填了「转场」，但助跑属于撑越跳马这唯一步骤本身，应为「操作中」。",
+    "n_issues": 1,
+    "n_major": 0
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、final=false、断点(操作中)、步态现象均与视频一致：机器人全程直立慢走，只走了约四分之一路程；根因留空并写了 gap_zh，合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:19Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-vault-i00-standard--vzeatet/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-vault-i00-standard--vzeatet/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-vault-i00-standard--vzeatet/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-vault-i00-standard--vzeatet/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "9892caa3015bf10fef88bae0861ff678cc94c2352b8c89e7c61a153818b56856",
+   "video_sha256": "6ae1fbc151b6859ddcf607dbe8dc17586df0af76bef32163de2a8b506bf70273",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}
diff --git a/phaseR/embodied-olympics-fun/volleyball_spike/trajectories/embodied-olympics-fun-volleyball-spike-i00-realtime--62syasc.json b/phaseR/embodied-olympics-fun/volleyball_spike/trajectories/embodied-olympics-fun-volleyball-spike-i00-realtime--62syasc.json
new file mode 100644
index 0000000000000000000000000000000000000000..040a8747e21278deaef342622866ef5e6a772a5e
--- /dev/null
+++ b/phaseR/embodied-olympics-fun/volleyball_spike/trajectories/embodied-olympics-fun-volleyball-spike-i00-realtime--62syasc.json
@@ -0,0 +1,271 @@
+{
+ "schema": "efe-trajectory-labels/1",
+ "taxonomy": "v2",
+ "runId": "embodied-olympics-fun-volleyball-spike-i00-realtime--62syasc",
+ "taskKey": "embodied-olympics-fun/volleyball_spike",
+ "dag_sha256": "f4bddfc185de81e8d53ee7f0cfb80e88756f3bb0aa176b76645d88cc0c11176b",
+ "outcome": {
+  "success": 0,
+  "reward": 0.0,
+  "source": "catalog",
+  "raw": {
+   "success": 0,
+   "live_success": 0,
+   "graded_by_live": 1,
+   "replays": 0,
+   "live_rule": 1,
+   "live_verdict_error": 0,
+   "finalized": 1,
+   "missing_log": 0,
+   "infra_error": 0,
+   "grader_error": 0,
+   "sim_error": 0,
+   "sim_crashed_on_call": 0,
+   "instance_mismatch": 0,
+   "n_actions": 412,
+   "sim_steps": 412,
+   "calls": 149,
+   "rows": 2306,
+   "resets": 0,
+   "reset_budget": 5,
+   "sim_startups": 1,
+   "elapsed_s": 4558.9,
+   "video_rendered": 1,
+   "first_success": 0,
+   "first_reason_code": 2,
+   "realtime": 1,
+   "slowdown": 1.0,
+   "clock_s": 5.142,
+   "max_lag_s": 0.5053,
+   "stalls": 2,
+   "stalled_s": 1.006,
+   "stalled_share": 0.1956,
+   "held_steps": 0,
+   "late_steps": 295,
+   "requests_refused": 3,
+   "request_busy_s": 0.691,
+   "resource_enforced": 0,
+   "limits_violation_observed": 0,
+   "resource_monitor_missing": 0,
+   "peak_gpu_memory_gb": 0.0,
+   "peak_memory_gb": 0.784,
+   "peak_processes": 9,
+   "peak_disk_gb": 0.436,
+   "peak_cpus": 1.58,
+   "peak_pids": 173,
+   "resource_memory_kills": 0,
+   "resource_pids_kills": 0,
+   "resource_disk_kills": 0,
+   "resource_gpu_memory_kills": 0,
+   "resource_monitor_kills": 0,
+   "resource_emergency_kills": 0,
+   "resource_limit_exceeded": 0,
+   "resource_cpu_throttles": 0,
+   "resource_cpu_throttled_sec": 0.0,
+   "resource_oom_kills": 0,
+   "resource_host_oom_kills": 0,
+   "resource_service_oom_kills": 0
+  }
+ },
+ "video": {
+  "kind": "media",
+  "url": "media:embodied-olympics-fun-volleyball-spike-i00-realtime--62syasc.mp4",
+  "duration_s": 4.16,
+  "fps": 25.0,
+  "frames": 104,
+  "sha256": "060f1b6eda05df861937f1e1e794c5f489bc6e73f75d85454b3606d8d22d6289"
+ },
+ "state": null,
+ "progress": {
+  "method": "visual",
+  "denominator": 1,
+  "base": 0,
+  "peak": 0,
+  "final": 0,
+  "nodes": {
+   "spike_in_court_landed": {
+    "method": "visual",
+    "intervals": [],
+    "final": false,
+    "evidence": [
+     {
+      "t": 3.4,
+      "log_step": 126,
+      "text_zh": "观察：视频 3.0-3.8 s 的斜俯视画面中，机器人站在网前球场边线外，没有可见腾空；球在画面右上方（视频约 3.2-3.4 s）飞过，未见拳垫击中球后越网落入对方橙色场区的过程。视频约 4.0 s 机器人倒地。画面小，触球瞬间无法确认。 观察：服务读数 outcome='the robot touched the net'、stages_reached 0/4（log 126）；球在 4.1 s 仍在下落、无折返，未被击中。"
+     }
+    ],
+    "contact_actual": [
+     {
+      "who": "未知",
+      "how": "未知",
+      "note_zh": "观察：画面分辨率低，看不清机器人腕端拳垫是否碰到球。"
+     }
+    ]
+   }
+  },
+  "constraints": {}
+ },
+ "native_metrics": [
+  {
+   "name": "contact_height_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 2.43,
+   "direction": "higher_better"
+  },
+  {
+   "name": "jump_rise_m",
+   "value": null,
+   "unit": "m",
+   "threshold": 0.1,
+   "direction": "higher_better"
+  },
+  {
+   "name": "speed_in_m_s",
+   "value": null,
+   "unit": "m/s",
+   "threshold": null,
+   "direction": null
+  },
+  {
+   "name": "speed_out_m_s",
+   "value": null,
+   "unit": "m/s",
+   "threshold": "speed_in_m_s",
+   "direction": "higher_better"
+  },
+  {
+   "name": "air_s",
+   "value": null,
+   "unit": "s",
+   "threshold": null,
+   "direction": null
+  },
+  {
+   "name": "landing_x_m",
+   "value": null,
+   "unit": "m",
+   "threshold": [
+    3.0,
+    12.0
+   ],
+   "direction": "target"
+  },
+  {
+   "name": "landing_y_m",
+   "value": null,
+   "unit": "m",
+   "threshold": [
+    -8.0,
+    1.0
+   ],
+   "direction": "target"
+  },
+  {
+   "name": "ball_height_m",
+   "value": null,
+   "unit": "m",
+   "threshold": null,
+   "direction": null
+  }
+ ],
+ "failure": {
+  "breakpoint": {
+   "ref": "spike_in_court_landed",
+   "where": "操作中",
+   "t": 3.4,
+   "evidence_zh": "观察：视频 3.0-4.16 s 中球未被可见地击入对方场区（球 4.1 s 仍沿原轨迹下落、无折返），机器人无明显腾空，约 3.5 s 起前倾、约 4.0 s 倒地；服务读数（log 126，评分 episode 自身）outcome='the robot touched the net'、stages_reached 0/4、control_steps 412（≈4.12 s）、reason failed；评分器 first_reason_code=2、success=0。",
+   "log_step": 126
+  },
+  "phenomena": [
+   {
+    "cat": "全身",
+    "item": "平衡",
+    "t": 3.9,
+    "log_step": 126,
+    "evidence_zh": "观察：视频约 3.5-4.16 s 机器人前倾、靠向球网并倒地；服务读数 outcome='the robot touched the net'（N5，评分 episode 自身读数，control_steps 412≈4.12 s）；球全程未被击中。"
+   }
+  ],
+  "root_causes": [],
+  "gap_zh": "视频仅 4.16 s、分辨率低，触球瞬间与球落点看不清；无状态日志，评分器未给出接触高度等字段。agent 的离线 MuJoCo 调参不算证据，但 log 126 的服务读数（outcome=触网、control_steps 412、stages_reached 0/4）按 N5 算观察。倒地的根因（log 125-126 显示整段 walk+jump 为开环预排队，可能缺姿态反馈）缺少直接证据，故根因留空。",
+  "unmet": {
+   "spike_in_court_landed": false
+  }
+ },
+ "summary_zh": "观察：唯一节点“扣球并落地站稳”未达成，视频 4.16 s 中球未被击中，机器人约 3.5 s 起前倾、约 4.0 s 倒地，服务读数为触网、stages_reached 0/4。推断：断点在击球步骤的操作中，现象为平衡；因无状态日志，根因证据不足而留空。",
+ "conf": 5,
+ "review": {
+  "status": "agreed",
+  "rounds": 2,
+  "reviewers": [
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 1,
+    "verdict": "fix",
+    "conf": 6,
+    "summary_zh": "节点值(未达成)、断点 ref 与 where 没问题；但 phenomena/root_causes 被留空，且 gap_zh 称“无状态日志、agent 日志非图中证据”不对：这是 live 单次评分 episode，服务读数(outcome=“the robot touched the net”、stages_reached=0)与视频倒地(≈3.5-4.16 s)足以支持“平衡”现象，应补上；证据里也漏了服务读数。",
+    "n_issues": 4,
+    "n_major": 1
+   },
+   {
+    "engine": "claude",
+    "model": "claude-sonnet-5-5",
+    "round": 2,
+    "verdict": "agree",
+    "conf": 7,
+    "summary_zh": "节点未达成、断点(操作中)、现象“平衡”与服务读数(触网、stages 0/4、412步≈4.12 s)均与视频一致；根因留空并有 gap_zh，合理。",
+    "n_issues": 0,
+    "n_major": 0
+   }
+  ],
+  "open_zh": []
+ },
+ "provenance": {
+  "prompt": "prompts/03_trajectory.md",
+  "prompt_sha256": "2385b35930667330fc4ea181a15ea0818b49cfeae0df1700b8b08c988e72ad8e",
+  "engine": "claude",
+  "model": "claude-sonnet-5-5",
+  "effort": "high",
+  "created": "2026-11-08T20:59:27Z",
+  "review_prompt": "prompts/04_trajectory_review.md",
+  "review_prompt_sha256": "9afbc4ac19b1e1cf03732c8961ec46a618d682ee229343f66e3316794b14f9ba",
+  "reconcile_prompt": "prompts/05_reconcile.md",
+  "reconcile_prompt_sha256": "8c827af3ae3f77d09229399fb41283c50a8f725302cbe3b263b8051213dde2b2",
+  "calls": [
+   {
+    "stage": "traj",
+    "round": 0,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-volleyball-spike-i00-realtime--62syasc/03_label"
+   },
+   {
+    "stage": "traj_review",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-volleyball-spike-i00-realtime--62syasc/04_review_r1"
+   },
+   {
+    "stage": "reconcile",
+    "round": 1,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-volleyball-spike-i00-realtime--62syasc/05_reconcile_r1"
+   },
+   {
+    "stage": "traj_review",
+    "round": 2,
+    "ok": true,
+    "space": "work/runs/embodied-olympics-fun-volleyball-spike-i00-realtime--62syasc/04_review_r2"
+   }
+  ],
+  "review_engine": "claude",
+  "review_model": "claude-sonnet-5-5",
+  "inputs": {
+   "run_json_sha256": "36fa0117d2997708d9a6a8ed683f8c7a0d71180a98dfb77e515aa640add098cd",
+   "video_sha256": "060f1b6eda05df861937f1e1e794c5f489bc6e73f75d85454b3606d8d22d6289",
+   "states_sha256": null,
+   "video_choice": "HF replay.mp4 status 404; local graded video"
+  },
+  "writer_note": "Writer and reviewer: Claude Sonnet 5.5 (claude-sonnet-5-5) in Claude Code, separate sessions; phaseR = counted attempts of the seven original agents that had no phase-2 label (reruns after 2026-11-06 and runs newer than the phase-2 catalog), labelled 2026-11-08 by the owner's decision. Their earlier labels were written by GPT-6.1 Sol in Codex. Same prompts (v3.2), same phase-2 task DAGs, same validation."
+ }
+}