claude-code-skills
Pass
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- EXTERNAL_DOWNLOADS (LOW): The skill directs the agent to use
WebFetchto retrieve aREADME.mdfrom an untrusted repository (gmh5225/awesome-skills). This is an external ingestion point for untrusted data. - COMMAND_EXECUTION (LOW): Instructs the use of
git clone,ls, andheadon user-scoped directories (~/.claude/skills/). These are standard for the skill's purpose but involve filesystem and network interaction. - REMOTE_CODE_EXECUTION (LOW): Facilitates the installation of third-party code via
git clonefrom arbitrary repositories. While common for development tools, it encourages the execution of unverified remote code. - PROMPT_INJECTION (LOW): (Indirect) The skill ingest untrusted data from an external source without explicit boundary markers or sanitization logic.
- Ingestion points:
WebFetchofhttps://raw.githubusercontent.com/gmh5225/awesome-skills/refs/heads/main/README.md(SKILL.md). - Boundary markers: Absent; the content is fetched directly into the context.
- Capability inventory:
git clone,ls,head(SKILL.md). - Sanitization: Absent.
Audit Metadata