ah-submit-code-review
Pass
Audited by Gen Agent Trust Hub on Mar 27, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns or security risks identified. The skill implements functional automation using standard developer tools.- [PROMPT_INJECTION]: The skill processes untrusted data from GitHub PRs and external review files, creating a surface for indirect prompt injection.
- Ingestion points: PR titles and bodies fetched via
gh pr view, and content from local review files. - Boundary markers: External content is interpolated into the review format without explicit boundary markers.
- Capability inventory: The skill uses
gh apito post comments and update PR states (REQUEST_CHANGES/APPROVE). - Sanitization: The implementation uses quoted heredocs (
<<'EOF') when piping JSON payloads to theghcommand, which prevents shell interpolation of potentially malicious injected content.
Audit Metadata