Smart Todo - AI智能代办管理

智能代办管理 Skill。用于管理日常任务和代办事项，支持优先级标记(P0-P1)、 状态跟踪、智能重复检测、上下文保存、定时提醒等功能。 当用户需要添加、查看、更新代办，或检测到工作中断时自动使用。

What to consider before installing: - Privacy: The skill scans your workspace for recently modified files and reads .workbuddy/memory/*.md to extract paths and conversation snippets. If your workspace contains secrets or private files, those could be captured into todo context. - Silent saving: SKILL.md explicitly says it will silently save context even if the user refuses creating a todo. If you want explicit consent control, edit SKILL.md or the code before enabling automatic triggers. - Storage location: assets/config.json defaults to ~/smart-todo-data. Change storage_path to a dedicated directory you control (preferably not a repo root or a directory containing secrets) before first run. - Audit & sandbox: Review scripts/context_capture.py and scripts/todo_manager.py line-by-line (they are included) or run the skill in a sandboxed environment first. Look for any places you’d like to disable (e.g., reading .workbuddy memory or os.walk behavior). - Configuration changes: Consider lowering or disabling automatic 'work interruption' triggers, or modify the flow so that nothing is saved unless the user explicitly confirms. Also limit how many files are read or remove heuristics that parse other markdown files. - Backup: Because the skill will create and overwrite active.md/archive.md, back up any existing files that might be overwritten. If you want, I can point out exact lines in the two Python files that implement the workspace scan, .workbuddy memory read, and the 'silent save' behavior so you can edit them safely.

Type: OpenClaw Skill Name: smart-todo Version: 1.0.0 The smart-todo skill is a task management system that uses local Markdown files (active.md, archive.md) to track user tasks. The included Python scripts (todo_manager.py and context_capture.py) perform legitimate functions such as calculating task similarity, managing task states, and gathering local context (recent file modifications and conversation history) to enrich todo entries. There is no evidence of data exfiltration, network activity, or malicious command execution; the behavior is entirely consistent with the stated purpose of a productivity assistant.