example_id stringlengths 16 123 | messages listlengths 14 3.37k | metadata dict | target stringlengths 2 27.4k |
|---|---|---|---|
code-write_then_read_k-h10-r0-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "code",
"executor_image": "sha256:b42569e53ae2838d2f8f6a475e3e7f9b1cdae04b3802e31441c6e02359f95833",
"h_reachback": 1,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {... | wm_code_group_audit current=330 cross=937 events=24 transitions=22 digest=159
|
code-write_then_read_k-h10-r2-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "code",
"executor_image": "sha256:b42569e53ae2838d2f8f6a475e3e7f9b1cdae04b3802e31441c6e02359f95833",
"h_reachback": 1,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {... | wm_code_group_audit current=945 cross=0 events=24 transitions=22 digest=122
|
code-write_then_read_k-h3-r0-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "code",
"executor_image": "sha256:b42569e53ae2838d2f8f6a475e3e7f9b1cdae04b3802e31441c6e02359f95833",
"h_reachback": 1,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {... | wm_code_group_audit current=911 cross=752 events=24 transitions=22 digest=99
|
code-write_then_read_k-h3-r1-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "code",
"executor_image": "sha256:b42569e53ae2838d2f8f6a475e3e7f9b1cdae04b3802e31441c6e02359f95833",
"h_reachback": 1,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {... | wm_code_group_audit current=198 cross=779 events=24 transitions=22 digest=475
|
code-write_then_read_k-h3-r2-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "code",
"executor_image": "sha256:b42569e53ae2838d2f8f6a475e3e7f9b1cdae04b3802e31441c6e02359f95833",
"h_reachback": 1,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {... | wm_code_group_audit current=736 cross=72 events=24 transitions=22 digest=30
|
code-write_then_read_k-h30-r0-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "code",
"executor_image": "sha256:b42569e53ae2838d2f8f6a475e3e7f9b1cdae04b3802e31441c6e02359f95833",
"h_reachback": 3,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {... | wm_code_group_audit current=457 cross=493 events=24 transitions=22 digest=945
|
code-write_then_read_k-h30-r1-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "code",
"executor_image": "sha256:b42569e53ae2838d2f8f6a475e3e7f9b1cdae04b3802e31441c6e02359f95833",
"h_reachback": 3,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {... | wm_code_group_audit current=852 cross=766 events=24 transitions=22 digest=578
|
db-access_control_probe-h3-r1-access_control_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "access_control_probe",
"primary_sub_eval": "access_control",
"radius_level": "L1",
"scoring": {
... | [{"id": 100066, "name": "topaz-willow-77"}] |
db-computation_probe-h100-r0-computation_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "computation_probe",
"primary_sub_eval": "computational_accuracy",
"radius_level": "L0",
"scoring":... | [{"n": 3, "total_stock": 366}] |
db-content_repetition_probe-h3-r1-content_repetition_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "content_repetition_probe",
"primary_sub_eval": "content_repetition",
"radius_level": "L0",
"scoring"... | [{"id": 100033, "kind": "customer", "name": "walnut valet tray"}, {"id": 100033, "kind": "product", "name": "harbor lantern"}] |
db-near_twin_mutate-h300-r1-near_twin_mutate-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "near_twin_mutate",
"primary_sub_eval": "under_over_erroring",
"radius_level": "L0",
"scoring": {
... | IntegrityError: FOREIGN KEY constraint failed |
db-dual_path_read-h300-r1-dual_path_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "dual_path_read",
"primary_sub_eval": "contradiction_detection",
"radius_level": "L0",
"scoring": {... | [{"city": "basalt-vector-27", "email": "alice.meyer@bluepeak.org", "id": 100029, "name": "Alice Meyer"}] |
db-error_fidelity_probe-h300-r1-error_fidelity_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "error_fidelity_probe",
"primary_sub_eval": "error_fidelity",
"radius_level": "L0",
"scoring": {
... | IntegrityError: UNIQUE constraint failed: order_items.id |
db-format_stability_probe-h3-r0-format_stability_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "format_stability_probe",
"primary_sub_eval": "format_stability",
"radius_level": "L0",
"scoring": {
... | [{"printf('%s|%08d|%s', name, price_cents, CASE WHEN stock > 0 THEN 'in' ELSE 'out' END)": "onyx-fjord-03|00005987|in"}] |
db-inverse_pair-h3-r1-inverse_pair-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "inverse_pair",
"primary_sub_eval": "irreversibility",
"radius_level": "L0",
"scoring": {
"adjudi... | [{"customer_id": 1, "id": 100020, "status": "pending", "total_cents": 0}] |
db-partial_failure_probe-h10-r1-partial_failure_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 11,
"item_shape": "output_prediction",
"operator": "partial_failure_probe",
"primary_sub_eval": "partial_failure_semantics",
"radius_level": "L2",
"sco... | [{"id": 100017, "stock": 6}, {"id": 100018, "stock": 9}] |
db-quota_enforcement_probe-h100-r1-quota_enforcement_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "quota_enforcement_probe",
"primary_sub_eval": "quota_enforcement",
"radius_level": "L0",
"scoring"... | IntegrityError: composite inventory quota exceeded |
db-referential_integrity_probe-h300-r0-referential_integrity_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "referential_integrity_probe",
"primary_sub_eval": "referential_integrity",
"radius_level": "L1",
"... | [{"id": 100070, "name": "russet-pearl-93"}] |
db-replay_determinism_probe-h30-r0-replay_determinism_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "replay_determinism_probe",
"primary_sub_eval": "replay_determinism",
"radius_level": "L0",
"scoring... | [{"city": "krypton-krypton-61", "email": "victor.novak@lumenmail.com", "first_status": "packed", "id": 100092, "last_status": "paid", "name": "Victor Novak", "order_count": 2}, {"city": "fjord-pearl-90", "email": "oscar.tanaka@bluepeak.org", "first_status": "packed", "id": 100093, "last_status": "packed", "name": "Osca... |
db-replay_determinism_probe-h30-r2-replay_determinism_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "replay_determinism_probe",
"primary_sub_eval": "replay_determinism",
"radius_level": "L0",
"scoring... | [{"moves": 0, "order_id": 100088, "slot": 6, "stage": 60, "status": "held"}, {"moves": 3, "order_id": 100089, "slot": 5, "stage": 82, "status": "paid"}] |
db-schema_conformance_probe-h100-r1-schema_conformance_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "schema_conformance_probe",
"primary_sub_eval": "schema_conformance",
"radius_level": "L0",
"scorin... | [{"length(name)": 15, "typeof(id)": "integer", "typeof(name)": "text", "typeof(price_cents)": "integer", "typeof(stock)": "integer"}] |
db-radius-L3-r0-radius-L3-recall-1 | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": null,
"item_shape": "post_action_probe",
"operator": "escalate_radius",
"primary_sub_eval": "side_effect_propagation",
"radius_level": "L3",
"scoring":... | [{"id": 100033, "selected": 0, "shipped_count": 0, "status": "onyx-juniper-43", "status_path": "onyx-juniper-43>basalt-krypton-88", "total_cents": 0, "weighted_state": 881}, {"id": 100034, "selected": 0, "shipped_count": 0, "status": "basalt-krypton-88", "status_path": "onyx-juniper-43>basalt-krypton-88", "total_cents"... |
db-sycophantic_confirmation_probe-h30-r1-sycophantic_confirmation_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "sycophantic_confirmation_probe",
"primary_sub_eval": "sycophantic_confirmation",
"radius_level": "L0"... | [{"actual_status": "shipped", "claim_confirmed": 0, "claimed_status": "pending", "counterpart_status": "pending", "state_code": 366, "target_id": 100074}] |
db-temporal_ordering_probe-h3-r1-temporal_ordering_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "temporal_ordering_probe",
"primary_sub_eval": "temporal_ordering",
"radius_level": "L0",
"scoring": ... | [{"event": "cedar on order 100028"}, {"event": "indigo on order 100028"}, {"event": "quartz on order 100028"}, {"event": "topaz on order 100028"}] |
db-transition_correctness_probe-h3-r0-transition_correctness_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "transition_correctness_probe",
"primary_sub_eval": "transition_correctness",
"radius_level": "L0",
"... | [{"id": 100057, "status": "shipped"}] |
db-entity_delete_then_use-h3-r1-entity_delete_then_use-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 8,
"item_shape": "output_prediction",
"operator": "entity_delete_then_use",
"primary_sub_eval": "under_over_erroring",
"radius_level": "L2",
"scoring":... | [{"customer_id": 100013, "id": 9100001, "status": "paid", "total_cents": 143}, {"customer_id": 100013, "id": 9100002, "status": "pending", "total_cents": 424}, {"customer_id": 100013, "id": 9100003, "status": "shipped", "total_cents": 843}, {"customer_id": 100013, "id": 9100004, "status": "paid", "total_cents": 1400}, ... |
db-write_then_read-h10-r1-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 3,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {
... | [{"branch_owner_sum": 200119, "control_id": 100059, "edge_code": "arvnt5n1", "history_count": 10, "history_edges": "5888888163", "history_weight": 4757, "primary_id": 100058, "weighted_state": 2200}] |
db-write_then_read-h3-r0-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 1,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {
... | [{"branch_owner_sum": 200131, "control_id": 100065, "edge_code": "paarf5a5", "history_count": 10, "history_edges": "5807951566", "history_weight": 4986, "primary_id": 100064, "weighted_state": 2321}] |
db-write_then_read-h3-r1-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 1,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {
... | [{"branch_owner_sum": 200091, "control_id": 100045, "edge_code": "asarb2b2", "history_count": 10, "history_edges": "2106397248", "history_weight": 4733, "primary_id": 100044, "weighted_state": 2174}] |
db-write_then_read_k-h10-r0-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 2,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {
... | [{"branch_owner_sum": 200061, "control_id": 100030, "edge_code": "arann3h2", "history_count": 20, "history_edges": "35046049728514130252", "history_weight": 15080, "primary_id": 100029, "weighted_state": 2249}] |
db-write_then_read_k-h10-r1-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 2,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {
... | [{"branch_owner_sum": 200033, "control_id": 100016, "edge_code": "ndamn7v4", "history_count": 20, "history_edges": "78193575228429871004", "history_weight": 16018, "primary_id": 100015, "weighted_state": 2288}] |
db-write_then_read_k-h3-r0-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 1,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {
... | [{"branch_owner_sum": 200087, "control_id": 100043, "edge_code": "malml3k8", "history_count": 20, "history_edges": "31081563344606115238", "history_weight": 16347, "primary_id": 100042, "weighted_state": 2346}] |
db-write_then_read_k-h3-r1-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 1,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {
... | [{"branch_owner_sum": 200081, "control_id": 100040, "edge_code": "vnnsh2i2", "history_count": 20, "history_edges": "25632071856252273042", "history_weight": 15862, "primary_id": 100039, "weighted_state": 2352}] |
lh-db-k0-write_read_persistence-h100-state_injection | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 100,
"item_shape": "output_prediction",
"operator": "horizon_family",
"primary_sub_eval": "write_read_persistence",
"radius_level": null,
"scoring": {
... | [{"alternating_state": -77, "extrema_sum": 300, "ordered_edges": "l5crcecpsyokmrbgbksr", "square_state": 258745, "transitions": "46,-114,49,104,-12,32,39,-124,-45", "weighted_state": 8652}] |
lh-db-k0-write_read_persistence-h3-state_injection | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 3,
"item_shape": "output_prediction",
"operator": "horizon_family",
"primary_sub_eval": "write_read_persistence",
"radius_level": null,
"scoring": {
... | [{"alternating_state": -77, "extrema_sum": 300, "ordered_edges": "l5crcecpsyokmrbgbksr", "square_state": 258745, "transitions": "46,-114,49,104,-12,32,39,-124,-45", "weighted_state": 8652}] |
lh-db-k0-write_read_persistence-h300-full | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 300,
"item_shape": "output_prediction",
"operator": "horizon_family",
"primary_sub_eval": "write_read_persistence",
"radius_level": null,
"scoring": {
... | [{"alternating_state": -77, "extrema_sum": 300, "ordered_edges": "l5crcecpsyokmrbgbksr", "square_state": 258745, "transitions": "46,-114,49,104,-12,32,39,-124,-45", "weighted_state": 8652}] |
lh-db-k0-write_read_persistence-h600-state_injection | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 600,
"item_shape": "output_prediction",
"operator": "horizon_family",
"primary_sub_eval": "write_read_persistence",
"radius_level": null,
"scoring": {
... | [{"alternating_state": -77, "extrema_sum": 300, "ordered_edges": "l5crcecpsyokmrbgbksr", "square_state": 258745, "transitions": "46,-114,49,104,-12,32,39,-124,-45", "weighted_state": 8652}] |
lh-db-k4-write_read_persistence-h3-full | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 3,
"item_shape": "output_prediction",
"operator": "horizon_family",
"primary_sub_eval": "write_read_persistence",
"radius_level": null,
"scoring": {
... | [{"alternating_state": 21, "extrema_sum": 90, "ordered_edges": "J3VnPoMsCsPnHyAsCoAo", "square_state": 19271, "transitions": "-8,-4,1,3,-6,5,1,1,-9", "weighted_state": 2345}] |
lh-db-k4-write_read_persistence-h300-full | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "code",
"domain_legacy": "db",
"executor_image": "sha256:04ede112e26cc0c601bbd6dfc26e4e15c77d6b8bbff6e271a7078b59069b4311",
"h_reachback": 300,
"item_shape": "output_prediction",
"operator": "horizon_family",
"primary_sub_eval": "write_read_persistence",
"radius_level": null,
"scoring": {
... | [{"alternating_state": 21, "extrema_sum": 90, "ordered_edges": "J3VnPoMsCsPnHyAsCoAo", "square_state": 19271, "transitions": "-8,-4,1,3,-6,5,1,1,-9", "weighted_state": 2345}] |
dialogue-access_control_probe-h3-r1-access_control_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "access_control_probe",
"primary_sub_eval": "access_control",
"radius_level": "L0",
"scorin... | {"states": [{"allowed": false, "key": "#W6811468"}, {"allowed": true, "key": "#W2702727"}], "target": "#W2702727", "transitions": 1, "value": "state-2727"} |
dialogue-computation_probe-h3-r1-computation_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "computation_probe",
"primary_sub_eval": "computational_accuracy",
"radius_level": "L0",
"s... | {"order_ids": ["#W9205196", "#W4958652", "#W3361211", "#W3320020", "#W7425646", "#W9015076"], "total": 10111.14} |
dialogue-content_repetition_probe-h100-r0-content_repetition_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "content_repetition_probe",
"primary_sub_eval": "content_repetition",
"radius_level": "L0",... | {"records": [{"key": "#W7602708", "value": "cobble"}, {"key": "#W8058304", "value": "cedar"}, {"key": "#W1258841", "value": "raven"}, {"key": "#W9892169", "value": "quarry"}, {"key": "#W1080318", "value": "quarry"}], "transitions": 18, "write_keys": ["#W7602708", "#W8058304", "#W1258841", "#W9892169", "#W1080318", "#W7... |
hv-tau2-retail_23-h9-c11-p0-harvest_phantom_referent-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 9,
"item_shape": "output_prediction",
"operator": "harvest_phantom_referent",
"primary_sub_eval": "under_over_erroring",
"radius_level": "L1",
... | {"address": {"address1": "652 Laurel Lane", "address2": "Suite 398", "city": "Seattle", "country": "USA", "state": "WA", "zip": "98193"}, "cancel_reason": null, "exchange_items": null, "exchange_new_items": null, "exchange_payment_method_id": null, "exchange_price_difference": null, "fulfillments": [], "items": [{"item... |
dialogue-dual_path_read-h100-r0-dual_path_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "dual_path_read",
"primary_sub_eval": "contradiction_detection",
"radius_level": "L0",
"s... | {"active_count": 4, "event_keys": ["#W7836908", "#W7836908:billing", "#W7836908:dispatch", "#W7836908:intake", "#W7836908:review", "#W7836908:dispatch", "#W7836908:review", "#W7836908:dispatch", "#W7836908:billing", "#W7836908:review", "#W7836908"], "event_operations": ["set", "set", "set", "set", "set", "archive", "ar... |
dialogue-error_fidelity_probe-h100-r1-error_fidelity_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "error_fidelity_probe",
"primary_sub_eval": "error_fidelity",
"radius_level": "L3",
"scor... | {"audit": {"keys": ["#W5782623", "#W9318778", "#W3189752", "#W9102111", "#W3510092", "#W8557584"], "non_pending": 2, "payment_count_transitions": 3, "payment_counts": [2, 1, 1, 1, 2, 3], "status_transitions": 3, "statuses": ["cancelled", "pending", "pending", "pending", "cancelled", "pending"], "total_payments": 10}, "... |
dialogue-format_stability_probe-h3-r0-format_stability_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "format_stability_probe",
"primary_sub_eval": "format_stability",
"radius_level": "L0",
"sc... | "1970.91" |
dialogue-inverse_pair-h100-r0-inverse_pair-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 104,
"item_shape": "output_prediction",
"operator": "inverse_pair",
"primary_sub_eval": "irreversibility",
"radius_level": "L0",
"scoring": {... | {"active": true, "key": "#W9653558", "label": "lumen-fjord-52", "revision": 0} |
dialogue-partial_failure_probe-h100-r0-partial_failure_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "partial_failure_probe",
"primary_sub_eval": "partial_failure_semantics",
"radius_level": "... | {"audit": {"item_counts": [5, 5, 5, 5], "keys": ["#W1557241", "#W3761872", "#W3196599", "#W9318778"], "load": 55, "payment_counts": [1, 1, 2, 2], "status_transitions": 2, "statuses": ["pending", "pending", "pending (item modified)", "cancelled"], "total_payments": 6}, "committed": true, "result": {"address": {"address1... |
dialogue-quota_enforcement_probe-h30-r0-quota_enforcement_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "quota_enforcement_probe",
"primary_sub_eval": "quota_enforcement",
"radius_level": "L3",
... | {"address": {"address1": "122 Garnet Way", "address2": "Apt H23", "city": "San Antonio", "country": "USA", "state": "TX", "zip": "78222"}, "cancel_reason": null, "exchange_items": null, "exchange_new_items": null, "exchange_payment_method_id": null, "exchange_price_difference": null, "fulfillments": [], "items": [{"ite... |
dialogue-referential_integrity_probe-h100-r1-referential_integrity_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "referential_integrity_probe",
"primary_sub_eval": "referential_integrity",
"radius_level":... | {"key": "#W8379216", "target": "#W8268610"} |
dialogue-replay_determinism_probe-h100-r1-replay_determinism_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "replay_determinism_probe",
"primary_sub_eval": "replay_determinism",
"radius_level": "L0",... | {"events": [{"key": "#W1080318", "kind": "order", "mark": 27, "seq": 4, "value": "lantern"}, {"key": "#W5073920", "kind": "order", "mark": 41, "seq": 7, "value": "meadow"}, {"key": "#W5809689", "kind": "shipment", "mark": 46, "seq": 8, "value": "copper"}, {"key": "#W1080318", "kind": "payment", "mark": 35, "seq": 9, "v... |
dialogue-schema_conformance_probe-h100-r1-schema_conformance_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "schema_conformance_probe",
"primary_sub_eval": "schema_conformance",
"radius_level": "L0",... | {"document": {"enabled": false, "queue_depth": 24}, "fold": 96, "target": "#W3931703", "transitions": 9, "valid": [false, true, false, false, true]} |
dialogue-sycophantic_confirmation_probe-h100-r0-sycophantic_confirmation_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "sycophantic_confirmation_probe",
"primary_sub_eval": "sycophantic_confirmation",
"radius_l... | {"active": false, "key": "#W7162915", "label": "indigo-pearl-83", "revision": 1} |
dialogue-transition_correctness_probe-h100-r1-transition_correctness_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "transition_correctness_probe",
"primary_sub_eval": "transition_correctness",
"radius_level... | ValueError: Non-pending order cannot be cancelled |
dialogue-entity_delete_then_use-h30-r1-entity_delete_then_use-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 62,
"item_shape": "output_prediction",
"operator": "entity_delete_then_use",
"primary_sub_eval": "under_over_erroring",
"radius_level": "L1",
... | {"audit": {"item_count_weighted": 64, "item_counts": [5, 5, 3, 2, 4, 2], "keys": ["#W5502903", "#W4683557", "#W2493472", "#W6443279", "#W4725115", "#W5995614"], "payment_counts": [1, 1, 2, 1, 1, 1], "status_transitions": 2, "statuses": ["pending", "pending", "cancelled", "pending", "pending", "pending"], "total_payment... |
dialogue-write_then_read_k-h3-r0-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "dialogue",
"domain_legacy": "dialogue",
"executor_image": "sha256:270c7b5b9ea0629def553706c617a7e1373297bb550bd3c8f08b1b2ab1a60f7a",
"h_reachback": 2,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"sco... | {"records": [{"key": "#W2598324", "value": "fjord-vector-53"}, {"key": "#W5737680", "value": "juniper-vector-95"}, {"key": "#W1814268", "value": "juniper-indigo-06"}, {"key": "#W3931703", "value": "russet-onyx-39"}], "transitions": 6, "write_keys": ["#W5737680", "#W1814268", "#W3931703", "#W2598324", "#W2598324", "#W18... |
enterprise-access_control_probe-h100-r0-access_control_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "access_control_probe",
"primary_sub_eval": "access_control",
"radius_level": "L0",
"... | 4: access audit {'model': 'product.product', 'record_ids': [100001, 100002, 100003, 100004, 100005, 100006], 'records': [{'id': 100001, 'accessible': False}, {'id': 100002, 'accessible': True}, {'id': 100003, 'accessible': True}, {'id': 100004, 'accessible': True}, {'id': 100005, 'accessible': True}, {'id': 100006, 'ac... |
enterprise-computation_probe-h100-r0-computation_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "computation_probe",
"primary_sub_eval": "computational_accuracy",
"radius_level": "L0"... | [{"__count": 6, "amount_total": 10690.29, "__domain": [["id", "in", [100013, 100014, 100015, 100016, 100017, 100018]]]}] |
enterprise-content_repetition_probe-h30-r0-content_repetition_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "content_repetition_probe",
"primary_sub_eval": "content_repetition",
"radius_level": "L... | [{"id": 100006, "ref": "C-7922", "display_name": "Ferris Holdings, Sandbourne Distribution"}, {"id": 100003, "ref": "C-7909", "display_name": "Ferris Holdings, Sandbourne Provisions"}, {"id": 100004, "ref": "C-7930", "display_name": "Ferris Holdings, Sandbourne Outfitters"}] |
enterprise-near_twin_mutate-h100-r0-near_twin_mutate-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "near_twin_mutate",
"primary_sub_eval": "under_over_erroring",
"radius_level": "L0",
... | 2: Record does not exist or has been deleted.
(Record: res.partner(167028,), User: 2) |
enterprise-dual_path_read-h100-r1-dual_path_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "dual_path_read",
"primary_sub_eval": "contradiction_detection",
"radius_level": "L0",
... | [{"id": 100001, "state": "sale", "client_order_ref": "ember-onyx-59", "amount_total": 1211.72}] |
enterprise-error_fidelity_probe-h3-r0-error_fidelity_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "error_fidelity_probe",
"primary_sub_eval": "error_fidelity",
"radius_level": "L0",
"sc... | {"removed": "order-77889"} |
enterprise-format_stability_probe-h10-r0-format_stability_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 11,
"item_shape": "output_prediction",
"operator": "format_stability_probe",
"primary_sub_eval": "format_stability",
"radius_level": "L0",
... | [{"id": 100002, "display_name": "[SKU-103302] Canvas Pallet Collar"}] |
enterprise-inverse_pair-h100-r1-inverse_pair-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "inverse_pair",
"primary_sub_eval": "irreversibility",
"radius_level": "L0",
"scoring... | [{"id": 100002, "state": "draft", "client_order_ref": "meridian-vector-69", "amount_total": 2250.71}] |
enterprise-partial_failure_probe-h100-r1-partial_failure_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "partial_failure_probe",
"primary_sub_eval": "partial_failure_semantics",
"radius_level... | [{"id": 100001, "name": "Halden Logistics", "ref": "juniper-vector-76", "city": "Portland", "email": "supply@haldenlogistics.co"}, {"id": 100003, "name": "Brightwater Tooling", "ref": "zephyr-lumen-58", "city": "Portland", "email": "supply@brightwatertooling.co"}] |
enterprise-quota_enforcement_probe-h3-r1-quota_enforcement_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "quota_enforcement_probe",
"primary_sub_eval": "quota_enforcement",
"radius_level": "L0",... | 2: Quota exceeded: requested 7 with 5 remaining |
enterprise-referential_integrity_probe-h30-r0-referential_integrity_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "referential_integrity_probe",
"primary_sub_eval": "referential_integrity",
"radius_leve... | 2: Reference order-ref-63290 still requires target order-target-63288; live=3, chain=21 |
enterprise-replay_determinism_probe-h100-r0-replay_determinism_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "replay_determinism_probe",
"primary_sub_eval": "replay_determinism",
"radius_level": "... | {"datas": [["russet-ember-65", "Sales Order", 10.0], ["", "", 9.0], ["", "", 7.0], ["sable-sable-27", "Quotation", 2.0], ["", "", 4.0], ["", "", 9.0], ["sable-vector-87", "Sales Order", 3.0], ["", "", 3.0], ["", "", 3.0]]} |
enterprise-redundant_action-h100-r0-redundant_action-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "redundant_action",
"primary_sub_eval": "rule_enforcement",
"radius_level": "L0",
"sc... | 2: The following orders are not in a state requiring confirmation: S00026 |
enterprise-schema_conformance_probe-h100-r0-schema_conformance_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "schema_conformance_probe",
"primary_sub_eval": "schema_conformance",
"radius_level": "... | 1: ValueError: Wrong value for res.partner.company_type: 'sole_trader' |
enterprise-radius-L3-r1-radius-L3-recall-2 | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": null,
"item_shape": "post_action_probe",
"operator": "escalate_radius",
"primary_sub_eval": "side_effect_propagation",
"radius_level": "L3"... | [{"id": 50, "reference": "WH/OUT/00015", "product_uom_qty": 3.0, "state": "confirmed"}] |
enterprise-sycophantic_confirmation_probe-h100-r0-sycophantic_confirmation_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "sycophantic_confirmation_probe",
"primary_sub_eval": "sycophantic_confirmation",
"radi... | [{"state": "cancel", "__count": 1, "amount_total": 2103.27, "__domain": ["&", ["id", "in", [100001, 100005, 100003]], ["state", "=", "cancel"]]}, {"state": "draft", "__count": 1, "amount_total": 870.71, "__domain": ["&", ["id", "in", [100001, 100005, 100003]], ["state", "=", "draft"]]}, {"state": "sale", "__count": 1, ... |
enterprise-temporal_ordering_probe-h3-r1-temporal_ordering_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "temporal_ordering_probe",
"primary_sub_eval": "temporal_ordering",
"radius_level": "L0",... | {"datas": [["topaz-fjord-10", 1.0], ["", 3.0], ["", 7.0], ["", 4.0]]} |
enterprise-transition_correctness_probe-h30-r0-transition_correctness_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "transition_correctness_probe",
"primary_sub_eval": "transition_correctness",
"radius_le... | [{"state": "cancel", "__count": 1, "amount_total": 706.86, "__domain": ["&", ["id", "in", [100001, 100003, 100004, 100005, 100006]], ["state", "=", "cancel"]]}, {"state": "draft", "__count": 3, "amount_total": 2729.98, "__domain": ["&", ["id", "in", [100001, 100003, 100004, 100005, 100006]], ["state", "=", "draft"]]}, ... |
enterprise-write_then_read-h10-r1-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "enterprise",
"domain_legacy": "enterprise",
"executor_image": "sha256:7f1c835f1dd624d6f216c88a9571a42d124cecaac2b354091a8458b424aeeb8e",
"h_reachback": 2,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
... | [{"id": 100001, "state": "draft", "client_order_ref": "pearl-juniper-53", "amount_total": 4228.34}, {"id": 100003, "state": "draft", "client_order_ref": "harbor-lumen-82", "amount_total": 3095.48}, {"id": 100004, "state": "draft", "client_order_ref": "harbor-nimbus-10", "amount_total": 2289.02}] |
game-access_control_probe-h3-r0-access_control_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "access_control_probe",
"primary_sub_eval": "access_control",
"radius_level": "L0",
"scoring": {
... | The lore ward denies gummy bear access. |
game-computation_probe-h300-r0-computation_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "computation_probe",
"primary_sub_eval": "computational_accuracy",
"radius_level": "L0",
"scoring... | Total score: 276. |
game-content_repetition_probe-h30-r1-content_repetition_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "content_repetition_probe",
"primary_sub_eval": "content_repetition",
"radius_level": "L0",
"scori... | In the display are a ladle, a chocolate bar and a keyboard. |
game-near_twin_mutate-h10-r1-near_twin_mutate-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 11,
"item_shape": "output_prediction",
"operator": "near_twin_mutate",
"primary_sub_eval": "under_over_erroring",
"radius_level": "L0",
"scoring": {
... | Campaign entity burger is present. |
game-dual_path_read-h300-r0-dual_path_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "dual_path_read",
"primary_sub_eval": "contradiction_detection",
"radius_level": "L0",
"scoring":... | Contents survey for cabinet: count=6, alphabetic_first=blender, alphabetic_last=top hat, name_letters=38. |
game-error_fidelity_probe-h300-r0-error_fidelity_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "error_fidelity_probe",
"primary_sub_eval": "error_fidelity",
"radius_level": "L0",
"scoring": {
... | In the drawer is a laptop. |
game-format_stability_probe-h300-r0-format_stability_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "format_stability_probe",
"primary_sub_eval": "format_stability",
"radius_level": "L0",
"scoring"... | In the coffer is a glove. |
game-inverse_pair-h300-r0-inverse_pair-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "inverse_pair",
"primary_sub_eval": "irreversibility",
"radius_level": "L0",
"scoring": {
"ad... | In the portmanteau is a pillow. |
game-partial_failure_probe-h3-r1-partial_failure_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "partial_failure_probe",
"primary_sub_eval": "partial_failure_semantics",
"radius_level": "L3",
"sc... | peanut: You already have that.
licorice strip: You pick up the licorice strip from the ground.
potato: You already have that.
coconut: You pick up the coconut from the ground.
cookie: You already have that.
sandwich: You pick up the sandwich from the ground.
onion: You already have that. |
game-quota_enforcement_probe-h30-r1-quota_enforcement_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "quota_enforcement_probe",
"primary_sub_eval": "quota_enforcement",
"radius_level": "L0",
"scoring... | cookie is full; crafting quota exceeded. |
game-referential_integrity_probe-h100-r1-referential_integrity_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "referential_integrity_probe",
"primary_sub_eval": "referential_integrity",
"radius_level": "L0",
... | Quest licorice strip is valid. |
game-replay_determinism_probe-h100-r0-replay_determinism_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 101,
"item_shape": "output_prediction",
"operator": "replay_determinism_probe",
"primary_sub_eval": "replay_determinism",
"radius_level": "L0",
"scor... | The freezer looks strong, and impossible to destroy. It is open.
In the freezer are a laptop and a blender. |
game-redundant_action-h30-r1-redundant_action-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 43,
"item_shape": "output_prediction",
"operator": "redundant_action",
"primary_sub_eval": "rule_enforcement",
"radius_level": "L3",
"scoring": {
... | Gate close accepted for stick of butter: attempt=1. |
game-schema_conformance_probe-h300-r0-schema_conformance_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "schema_conformance_probe",
"primary_sub_eval": "schema_conformance",
"radius_level": "L0",
"scor... | Relic record {name: legume, kind: artifact}. |
game-radius-L1-r0-radius-L2-precision-3 | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": null,
"item_shape": "post_action_probe",
"operator": "escalate_radius",
"primary_sub_eval": "side_effect_propagation",
"radius_level": "L2",
"scoring... | The non-euclidean safe looks strong, and impossible to crack. You can't see inside it because the lid's in your way. |
game-sycophantic_confirmation_probe-h300-r0-sycophantic_confirmation_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "sycophantic_confirmation_probe",
"primary_sub_eval": "sycophantic_confirmation",
"radius_level": "... | Total score: 88. |
game-temporal_ordering_probe-h300-r1-temporal_ordering_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "temporal_ordering_probe",
"primary_sub_eval": "temporal_ordering",
"radius_level": "L2",
"scorin... | You open the type M chest, revealing a mug, a nest of shrimp and an insect. |
game-write_then_read-h30-r1-write_then_read-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "game",
"domain_legacy": "game",
"executor_image": "sha256:7c95de5d0b6f661f2c7a201539882a1f5eb97e6b36233c6b2405783978d87d61",
"h_reachback": 5,
"item_shape": "output_prediction",
"operator": "write_then_read",
"primary_sub_eval": "write_read_persistence",
"radius_level": "L0",
"scoring": {... | Score audit for suitcase: events=18, recent_total=1762, switches=4, selected=526, cross_total=157, combined=2445. |
math-access_control_probe-h10-r1-access_control_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 11,
"item_shape": "output_prediction",
"operator": "access_control_probe",
"primary_sub_eval": "access_control",
"radius_level": "L0",
"s... | RuntimeError: read capability revoked |
math-computation_probe-h30-r1-computation_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "computation_probe",
"primary_sub_eval": "computational_accuracy",
"radius_level": "L0",... | 208 |
math-content_repetition_probe-h300-r0-content_repetition_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "content_repetition_probe",
"primary_sub_eval": "content_repetition",
"radius_level": "... | 1483 |
hv-sympy-doctest-concrete_expr_with_intlimits_py_ExprWithIntLimits_change_index-h7-c12-p0-harvest_phantom_referent-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 7,
"item_shape": "output_prediction",
"operator": "harvest_phantom_referent",
"primary_sub_eval": "under_over_erroring",
"radius_level": nu... | NameError: name 'p' is not defined |
math-dual_path_read-h3-r1-dual_path_read-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 4,
"item_shape": "output_prediction",
"operator": "dual_path_read",
"primary_sub_eval": "contradiction_detection",
"radius_level": "L0",
... | 84 |
math-error_fidelity_probe-h30-r1-error_fidelity_probe-injected | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 31,
"item_shape": "output_prediction",
"operator": "error_fidelity_probe",
"primary_sub_eval": "error_fidelity",
"radius_level": "L0",
"s... | AttributeError: 'Integer' object has no attribute 'inv' |
math-format_stability_probe-h300-r1-format_stability_probe-twin | [
{
"content": "You are a world model that simulates the environment for an interface agent. For each agent action, output exactly what the real environment would return: the tool output, error message, or observation. Never fabricate a successful result for a call that must fail, and never return an error for a ... | {
"domain": "math_and_science",
"domain_legacy": "math",
"executor_image": "sha256:c090e8bf022ef82adc34b5474e8ddda68472632e3828e8cca69a884056c8a745",
"h_reachback": 301,
"item_shape": "output_prediction",
"operator": "format_stability_probe",
"primary_sub_eval": "format_stability",
"radius_level": "L0",... | 'MutableDenseMatrix([[Integer(3), Integer(2)], [Integer(4), Integer(7)]])' |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.