A Feynman-based learning coach for systematic multi-day learning plans and quick article/book breakdowns. Supports EN and ZH.
Security Analysis
high confidenceThe skill's requests and included code are coherent with a learning-coach: it generates and stores learning plans, fetches web sources, and can schedule daily pushes — it does not ask for unrelated credentials or install external code.
Name/description (Feynman-based learning coach) match the behavior in SKILL.md and the included script: generating learning paths, quizzes, source-search guidance, saving plan files under learning-data/{slug}, and scheduling daily pushes. No unrelated credentials, binaries, or config paths are requested.
The runtime instructions direct the agent to perform web searches, fetch and inspect pages (headers, dates, GitHub commit times), write/read files under learning-data/{slug}, and create cron-style scheduled pushes. All of these actions are appropriate for a learning coach, but they do involve network fetches and persistent storage of plan/state data.
No install spec; this is instruction-only plus a small helper script (scripts/learner.py) included in the bundle. There are no downloads from external URLs or package installs, so nothing is written to disk beyond the skill's own files and any learning-data the agent creates during use.
The skill requires no environment variables, no credentials, and no config paths. The actions described (web fetching and local plan files) do not require additional secrets; requested scope is proportionate to the described functionality.
The skill instructs creation of per-plan cron schedules and persistent files (learning-data/{slug}/path.json, review-state.json, etc.). always is false. Scheduling recurring pushes is consistent with a learning coach but is a persistence capability users should be aware of (it will schedule future agent actions and write plan state to disk).
Guidance
This skill appears internally consistent with its stated purpose. Before installing, be aware that: (1) it will create and store plan/state files under learning-data/{slug} on the agent host; (2) it instructs the agent to perform web searches and fetch pages (network activity) to verify sources; (3) it will create scheduled cron-like jobs to push daily content — these recurring actions persist until disabled. The included script (scripts/learner.py) only generates path/quiz JSON and lists plans. If you accept scheduled pushes and local storage for plans, this skill is reasonable to use. If you are uncomfortable with autonomous scheduled pushes or persistent files, review/inspect the skill files and disable autonomous invocation or refuse cron creation before enabling the skill.
Latest Release
v2.5.0
Language policy optimization: allow natural mixing of technical terms (State, Props, useEffect) in user's primary language, while forbidding gratuitous mixing. All reference files updated with consistent language notes.
More by @wjyhahaha
Published by @wjyhahaha on ClawHub
