receiving-code-review
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external input (code review feedback) which may influence the agent's behavior or code modifications.
- Ingestion points: The skill explicitly handles feedback from 'External Reviewers' and 'human partners' as outlined in
SKILL.md. - Boundary markers: The instructions lack specific delimiters or technical warnings to isolate external feedback from the agent's core instructions, relying instead on the agent's logic to 'Verify before implementing'.
- Capability inventory: The skill utilizes tools for searching the codebase (
grep), interacting with the GitHub API (gh api), and carries an implicit capability to modify files when implementing fixes. - Sanitization: No programmatic sanitization or filtering of the feedback content is defined; the skill relies on the agent's technical rigor and 'skepticism' as a manual safeguard.
Audit Metadata