minimal-fix
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill does not contain any malicious patterns and actively implements security best practices:
- Path Denylist: Explicitly forbids editing sensitive files and directories such as
.env,auth/,payments/, and secrets. - Scope Limitation: Restricts the agent to the "smallest possible code change" and forbids unrelated refactoring, which limits the potential blast radius of any modification.
- Verification Requirements: Mandates reproduction of failures and execution of tests/linting to verify changes.
- Operational Safety: Recommends worktree isolation for unattended runs and explicitly states that the agent cannot mark its own work as done, ensuring a human or verifier remains in the loop.
Audit Metadata