dogfood-cli
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze an external codebase (vigiles), which acts as an untrusted data source. This creates a surface where malicious instructions embedded in the code could influence the agent's behavior during the fix and commit process.
- Ingestion points: The agent is directed to read and audit various distinct surfaces of the codebase, including compile, eject, and config files (as detailed in SKILL.md).
- Boundary markers: There are no explicit instructions or delimiters provided to help the agent distinguish its primary instructions from data/code it reads from the repository.
- Capability inventory: The agent is granted the ability to modify source files, execute shell commands (npm run build, vitest), and perform git operations (git commit).
- Sanitization: The skill lacks requirements for sanitizing or validating the contents of the audited code before the agent performs build or test operations.
Audit Metadata