copilot-review-loop
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes automated tests (identified from project configuration like
CLAUDE.md) to verify that applied code fixes do not break existing functionality. This is a standard development practice. - [PROMPT_INJECTION]: The skill processes external data (source code and review feedback) to suggest and apply code changes, creating a surface for indirect prompt injection. This risk is mitigated by the workflow's requirement for explicit user approval before any code modifications are applied ('apply_fixes' follows 'ask_user_which_to_address').
- [DATA_EXPOSURE]: The tool accesses local source files and git state (via
git diff --staged) to determine the scope of the review. This access is necessary for the skill's primary function and remains within the local environment.
Audit Metadata