review
Pass
Audited by Gen Agent Trust Hub on Aug 8, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions define a standard software engineering review process. No malicious patterns or behaviors were identified in the skill body, metadata, or associated incident logs. The vendor resources mentioned (e.g., fallow-core) are consistent with the skill's author.
- [PROMPT_INJECTION]: The skill includes an indirect prompt injection surface as it is designed to ingest and process untrusted code from pull requests and branch diffs. This is an inherent risk for review-oriented tasks, but the instructions include defensive measures like manual verification of source and live output rather than trusting automated results.
- Ingestion points: Code diffs, branch source code, and plan files referenced in SKILL.md.
- Boundary markers: Absent; the instructions do not specify the use of delimiters or 'ignore' instructions for separating untrusted data from the agent's processing logic.
- Capability inventory: The agent is directed to execute local repository quality gates and smoke tests, which may involve running scripts like 'npm run fmt:js:check' as mentioned in INCIDENT_LOG.md.
- Sanitization: Absent; the skill does not describe methods for filtering or escaping content from external pull requests before review.
Audit Metadata