code-review-playbook
Pass
Audited by Gen Agent Trust Hub on Jul 28, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill incorporates a robust defense against indirect prompt injection by defining a structured data processing workflow. 1. Ingestion points: PR descriptions and comments are identified as external data sources in Phase 1. 2. Boundary markers: The skill enforces a 'HARD-GATE' where the diff is the sole authority and PR text is explicitly labeled as untrusted. 3. Capability inventory: The skill defines no dangerous capabilities such as subprocess execution or network operations. 4. Sanitization: It requires that facts extracted from prose be verified against the code diff before being used in the analysis.
- [SAFE]: No other security concerns were identified, including obfuscation, credential exposure, or remote code execution.
Audit Metadata