pr-review
Pass
Audited by Gen Agent Trust Hub on May 16, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- Localized Script Execution: The skill invokes a PowerShell helper (collect-diff.ps1) to programmatically access repository state and code diffs. This is a controlled execution of internal logic required to provide context for the review.
- Review of External Content: The skill processes code and documentation from pull requests. It manages the potential for unexpected instructions in these diffs by using a structured output contract and isolating analysis within specific sub-agent tasks.
- Multi-Model Validation: A quality-focused pattern is used where a secondary model family cross-checks critical findings. This reduces the risk of automated hallucinations and ensures that the feedback provided to developers is more reliable.
Audit Metadata