Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name/description (deep content decoding, audit, visual analysis) aligns with requesting a target URL and performing text/visual analysis. The included safety_audit function and platform detection are coherent with that aim. However the code implements only a lightweight framework (returns a JSON analysis template) and does not fully implement the described features (see instruction_scope).
Instruction Scope
SKILL.md instructs the agent to fetch page metadata via agent-browser, run Python logic to perform text desensitization and produce a structured Markdown report. The shipped main.py, when executed, only accepts a URL and returns a JSON framework; it does not call safety_audit, does not accept or process fetched page content, and does not produce Markdown. This mismatch means runtime behavior may differ from user expectations and the instructions' stated data handling.
Install Mechanism
No install spec (instruction-only skill with a small helper script). Nothing is written to disk by an installer and no external downloads are requested.
Credentials
No environment variables, credentials, or config paths are requested. The skill does not require elevated access or unrelated secrets.
Persistence & Privilege
always is false and the skill does not request permanent presence or elevated platform privileges. Autonomous invocation is permitted by default (platform default) but not combined with other red flags.
What to consider before installing
This skill appears to aim at analyzing URLs and producing a decomposition template, but SKILL.md promises behavior (desensitization, full analysis, and Markdown output) that the included main.py does not implement. Before installing: (1) confirm how the agent-browser integration supplies page content to the skill — main.py currently only reads a URL and never consumes page text; (2) ask the author to reconcile SKILL.md and main.py (either have the code accept and sanitize fetched content, or update the docs to reflect that the skill only returns an analysis framework); (3) be aware that the '网安级审计' feature describes detecting moderation-evasion techniques — this capability can be dual-use, so ensure you are comfortable with the policy/legal implications of using a tool that analyzes how content may bypass platform filters; (4) because behavior may not match expectations, avoid enabling autonomous invocation for sensitive contexts until the inconsistencies are resolved.Like a lobster shell, security has layers — review code before you run it.
latestvk972gmspey4ysxepn2dnzakf7n84nrkc
License
MIT-0
Free to use, modify, and redistribute. No attribution required.