Catch
npx -y skills add CoralLips/thinking-tree --skill catchAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Manually capture missed insights from recent conversation. Use when thinking-tree's auto-recording missed something valuable. Accepts a natural language hint about what to capture.
SKILL.md
1.9 KB, as published. Nobody here has run it
补录捕获
当自动记录遗漏了有价值的对话认知时,用户可以用这个命令手动补录。
用法
/catch 我们刚才关于碎片转化速度的讨论
/catch 上面 AI 分析的那个架构边界的判断
/catch 最近几轮关于捕获和思考关系的讨论
$ARGUMENTS 是自然语言提示,描述想要捕获的内容方向。
执行步骤
1. 回溯对话
回顾最近 10-15 轮对话内容(当前会话内),重点关注:
- 用户提示方向(
$ARGUMENTS)相关的讨论 - AI 回复中的分析结论、架构判断、独立洞察
- 对话碰撞中产生的新认知
2. 提取认知
从回溯的对话中提取值得记录的独立认知点。每个点必须:
- 脱离对话上下文后仍然可理解
- 不与
~/.thinking-tree/fragments.md中已有碎片重复 - 有明确的一个点(不是模糊的"讨论了 XXX")
3. 路由判断
对每个提取出的认知点:
- 独立观点 → 碎片(fragment)
- 明确的疑问 → 问题(question)
- 具体可执行 → 行动项(todo)
4. 写入
通过原子写入脚本逐条写入:
echo '{"type":"fragment","title":"#标签 标题(日期)","body":"内容描述。"}' | node ~/.thinking-tree/bin/write-item.js
5. 输出确认
列出所有补录的条目:
补录完成:
📝 #标签1 标题1
📝 #标签2 标题2
❓ 问题标题
共 N 条
注意事项
- 先读
~/.thinking-tree/fragments.md确认不重复 - 碎片标题加
#标签前缀和日期后缀,与现有碎片风格一致 - 如果
$ARGUMENTS为空,扫描最近 5 轮对话,自动识别遗漏 - 每次补录不超过 5 条,避免噪音
- 写入使用
write-item.js,不要直接 Edit/Write 文件