awesome-code-review-feedback
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves processing untrusted external data from PR comments, which presents a surface for indirect prompt injection.
- Ingestion points: Ingests code review comments, bot summaries, and inline suggestions from external contributors or automated tools.
- Boundary markers: Employs explicit boundary instructions, stating that reviewer text must be treated as untrusted data and hints, not as authoritative instructions.
- Capability inventory: The skill facilitates codebase modification (implementation) and test execution (verification) as part of the feedback loop.
- Sanitization: Includes mandatory checks to ignore embedded instructions that ask to read secrets, modify auth files, or execute shell commands.
- [SAFE]: The skill is entirely instructional and does not ship with executable code or scripts. No patterns of obfuscation, exfiltration, or unauthorized privilege escalation were detected.
Audit Metadata