add-skill
Pass
Audited by Gen Agent Trust Hub on May 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/fetch_skill.pyusessubprocess.runto execute Git operations includingclone,sparse-checkout, andcheckout. These are called with argument lists, which prevents shell injection. - [EXTERNAL_DOWNLOADS]: The skill downloads directory contents from external GitHub repositories provided by the user.
- [PROMPT_INJECTION]: The skill serves as a delivery mechanism for indirect prompt injection, as it allows the agent to download and install external
SKILL.mdinstruction files. - Ingestion points: Files downloaded from external GitHub repositories.
- Boundary markers: None identified.
- Capability inventory:
subprocess.runfor system commands,shutilfor file system manipulation. - Sanitization: The script verifies the presence of
SKILL.mdbut does not validate or sanitize its content before installation.
Audit Metadata