!
Purpose & Capability
The skill is described as prompt-only and declares no required config paths or credentials, yet handler.py attempts to read a SKILL.md from a hard-coded local path (/Users/jianghaidong/.openclaw/skills/{skill_name}/SKILL.md). That local filesystem access is not justified by the stated purpose and is incoherent with the declared requirements.
!
Instruction Scope
SKILL.md contains only prompt guidance (no file/system access). The runtime code (handler.py) contradicts this by reading a local file. The README/instructions do not mention reading local files or using skill_name to load local content, so the instructions and the implementation are out of sync.
✓
Install Mechanism
No install spec, no external downloads, and no required binaries — low install risk. The only code is small and included in the package.
!
Credentials
The package declares no environment/credential needs, but the code accesses a specific user home path. Accessing arbitrary files in a user's home directory is disproportionate to the stated function and could expose local data. The code also takes a skill_name input and interpolates it into a filesystem path without sanitization (possible directory traversal).
✓
Persistence & Privilege
The skill is not always-enabled and does not request elevated privileges. It also does not declare persistence. However, local file reads still increase its blast radius compared to an instruction-only skill.
What to consider before installing
This skill's documentation says it's prompt-only and needs nothing, but the included handler.py tries to read /Users/jianghaidong/.openclaw/skills/{skill_name}/SKILL.md using an unsanitized skill_name. That mismatch is suspicious — it may be an accidental leftover from development, but it also allows local file reads (and potentially directory traversal) which could leak sensitive files. Before installing or enabling: 1) Ask the publisher why the code reads a hard-coded home path and request removal of any local-file reads unless explicitly needed. 2) If you or a reviewer will run the code, inspect/modify handler.py to (a) avoid hard-coded paths, (b) sanitize/validate skill_name (reject path separators), or (c) remove _load_skill_meta entirely and use the packaged SKILL.md via safe APIs. 3) Run the skill in a sandboxed environment if you must test it, and verify there are no network calls and it doesn't read sensitive files. 4) If you can't get a satisfactory explanation or patch, treat the skill as untrusted and avoid installing it in environments with sensitive data. The current state is likely a coding oversight but is not justified by the skill's description, so proceed with caution.
