dig
Pass
Audited by Gen Agent Trust Hub on Mar 3, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches library source code from GitHub repositories to provide documentation and usage insights.
- [COMMAND_EXECUTION]: Uses shell commands including
mkdir,ls, andgit cloneto manage repository files in the/tmp/cc-repos/directory. - [PROMPT_INJECTION]: The skill exhibits vulnerability to indirect prompt injection by processing untrusted external content.
- Ingestion points: Clones arbitrary repository data from GitHub into the local filesystem (SKILL.md).
- Boundary markers: The prompt for the research subagent does not include delimiters or instructions to ignore instructions embedded in the source code or documentation.
- Capability inventory: Includes shell command execution (
git clone,ls) and deep file exploration via a subagent with read access to all cloned files. - Sanitization: No sanitization or verification of the repository content is performed before the data is analyzed by the subagent.
Audit Metadata