codex-review-commit-message
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data, specifically the contents of
COMMIT_AGENTMSGand the output of git commands (diffs and logs), which could contain malicious instructions designed to influence the agent's behavior. - Ingestion points: Reads commit message text provided as
COMMIT_AGENTMSGand repository history/staged changes via git. - Boundary markers: The instructions lack explicit structural delimiters for the external data, though they do include a directive not to trust the caller's summary.
- Capability inventory: Limited to read-only git operations. The skill explicitly forbids editing files, staging paths, or running mutating commands.
- Sanitization: No specific sanitization or filtering of the ingested commit message or diff content is defined.
Audit Metadata