Jun 2 15:44 feat: Agent Team Jun 1 07:16 feat: SubAgent Swarm Topology May 24 03:42 feat: subagent inherit parent class ❯ 你尝试 调用 SubAgent 看能不能用
● 好的,我來測試一下 SubAgent(即 Agent tool)的功能。
● Agent(Test SubAgent functionality)
● SubAgent 正常運作 ✅
它成功執行了一個 general-purpose 類型的 agent,獨立完成了:
1. 列出專案目錄結構
2. 查看最近 5 筆 git commit
3. 回報結果給我
我再測試一下背景模式(background):
● Agent(Test background SubAgent)
⎿ Backgrounded agent (↓ to manage · ctrl+o to expand)
● SubAgent 兩種模式都正常運作:
┌───────────────────┬─────────────────────────────────┐
│ 模式 │ 狀態 │
├───────────────────┼─────────────────────────────────┤
│ Foreground (前景) │ ✅ 成功執行並回傳結果 │
├───────────────────┼─────────────────────────────────┤
│ Background (背景) │ ✅ 成功啟動,完成後會自動通知我 │
└───────────────────┴─────────────────────────────────┘
功能總結:
- 可以啟動多種 agent 類型(general-purpose、Explore、Plan 等)
- 支援前景(等結果)與背景(非同步)模式
- 可以同時啟動多個平行執行
有什麼具體任務想讓 SubAgent 幫忙處理的嗎?
✻ Cooked for 30s · 1 local agent still running
● Agent "Test background SubAgent" completed
● 背景 SubAgent 也成功完成了 ✅
Background age… May 23 16:49 feat: desktop cd desktop && bun run tauri dev 11:32 feat: add opencode zen provider /login and enter opencode zen free then /model input free chose you want May 21 17:08 chore: fix prompt identity 16:32 fix: opencode login May 7 16:44 fix: after /login opencode freemodel exit then /model can't fetch model list 16:27 fix: subagent can't created in opencode free model /login no apikey May 6 15:57 feat: better collapsing fetch result when set verbose is false in /config 13:19 feat: keep ctx line online 12:49 fix: move ctx line to rightdown 12:11 feat: ctx line 11:30 fix: /model display name not use fallback 10:12 fix: Not logged in . Run /login in opencode 10:05 chore: add free gpt5 nano and big pickle in /model 09:41 fix: refresh model list 09:14 fix: model not display life-free model 09:00 fix: fetch active opencode model 07:20 feat: add new free provider opencode Apr 30 15:07 feat: add Tavily as optional search backend in WebSearchTool Tavily PR Agent
Apr 21 13:43 fix: VersperClaw resume when ctrl c Apr 18 14:58 feat: toolsearch use websearch Apr 15 09:47 feat: auto dream ✅ Auto Dream 功能实现完成!
📋 实施总结
新增文件
- `src/skills/bundled/dream.ts` - 手动 /dream 命令实现
- 无条件可用(只要 auto-memory 启用)
- 完整工具权限,用户可实时观察
- 支持额外参数传递
修改文件
- `src/skills/bundled/index.ts` - 移除 KAIROS feature gate
- 添加 registerDreamSkill 导入
- 无条件调用 registerDreamSkill()
🎯 功能特性
自动触发(Auto Dream)
- ✅ 每次查询循环结束后检查
- ✅ 三重门控机制:
1. 启用状态检查
2. 时间门(默认 24 小时)
3. 会话门(默认 5 个会话)
4. 锁文件保护
- ✅ 后台 forked agent 运行,工具受限
- ✅ UI 中显示状态和进度
手动触发(/dream 命令)
- ✅ 用户随时可用,无门控限制
- ✅ 主循环中运行,完整工具权限
- ✅ 实时观察操作过程
- ✅ 支持额外上下文参数
4 阶段整理流程
1. Orient - 扫描现有记忆文件
2. Gather - 收集新信息(日志、过时记忆、会话记录)
3. Consolidate - 合并到现有文件
4. Prune and index - 更新索引并清理
🔧 配置选项
Settings.json:
{
"autoDreamEnabled": true,
"autoMemoryEnabled": true
}
GrowthBook:
- Feature: tengu_onyx_plover
- Parameters: minHours, minSessions
📊 与 BestClaude 对比
┌──────────────────┬────────────┬──────────┬──────────┐
│ 功能 │ Bes… 09:13 fix: rm Number(chatId) and use string save data project beyond chatId range Apr 13 02:18 feat: allow all tools in auto mode without classifier checking Modify isAutoModeAllowlistedTool to always return true, allowing all
tools to bypass the auto mode classifier and execute immediately. 02:03 feat: support auto-compact for "exceeds the available context size" errors Add error handler for alternative context overflow format used by some
API providers. This ensures auto-compact triggers regardless of the
specific error wording. 01:58 chore: back to clawd logo 01:10 feat: enable ToolSearch by default for all API providers Remove the restriction that disabled ToolSearch for non-first-party
APIs. This improves compatibility with OpenRouter and other third-party
providers, ensuring that deferred tools like TaskCreate work correctly
regardless of the API provider being used. 00:51 feat: load all openrouter model Apr 12 13:34 feat: live refresh typing status 13:02 chore: rm log function of webfetch 13:01 feat: complete shouldDefer tool WebFetch Apr 9 13:57 chore: original design to avoid ip issues Apr 8 13:24 chore: clean WebSearchTool log test 06:48 fix: WebSearch return data. 06:33 test: WebSearchTool log 06:13 fix: ban telegram log 06:05 test: WebSearchTool log 05:31 feat: WebSearchTool 05:09 chore: clean 02:38 feat: jina websearch and webfetch Apr 7 16:39 feat: config, jina fetch and search 06:14 fix: web saerch use python 02:46 feat: tls make web_search ok 主要改进
1. WebSearch 工具
- 集成 @yukiakai/tls-fetch 包,使用真实浏览器 TLS 指纹绕过
DuckDuckGo 的 CAPTCHA
- 修正 HTML 解析模式(class="...web-result...")
- 改进 URL 解码和链接提取逻辑
- 添加详细的调试日志
2. WebFetch 工具
- 实现指数退避重试机制
- 改进 Jina Reader 错误处理(检测 429 错误)
- 优化双层回退策略(Jina Reader → 直接抓取)
- 增强错误处理和日志
3. 依赖管理
- 添加 @yukiakai/tls-fetch 依赖(v2.0.2)
- 重新安装原生绑定解决兼容性问题
测试结果
✅ WebSearch 功能正常:
- 成功绕过 CAPTCHA 检测
- 正确解析搜索结果
- 支持中文和英文查询
- 返回完整的标题、URL 和摘要
✅ CLI 构建成功:
- 所有模块正确打包
- 原生模块加载正常
- 无运行时错误
修改的文件
- src/tools/WebSearchTool/WebSearchTool.ts
- src/tools/WebSearchTool/prompt.ts
- src/tools/WebFetchTool/WebFetchTool.ts
- src/tools/WebFetchTool/utils.ts
- package.json
- bun.lock Apr 6 18:57 fix: web search and fetch 16:05 feat: zero-search 14:59 fix: WebSearch API error 14:08 chore: WebFetchTool note 13:35 feat: open ripgrep with USE_BUILTIN_RIPGREP = '0'