technical-deslop
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill incorporates explicit safety instructions in references/slop-patterns.md and references/playbook.md, specifically mandating the preservation of security boundaries, permission checks, and input validation during the cleanup process.
- [SAFE]: The use of git commands like git diff is consistent with the skill's stated purpose of reviewing and cleaning code changes.
- [SAFE]: Analysis of indirect prompt injection risks (Category 8):
- Ingestion points: Untrusted code enters the agent context via git diff output as described in SKILL.md and references/playbook.md.
- Boundary markers: Not explicitly defined in the prompts.
- Capability inventory: The agent is empowered to modify local files based on its analysis of the diff.
- Sanitization: No formal sanitization is present, but the instructions include a 'Final review pass' to ensure no accidental API or logic changes occur. Given the primary purpose is code editing, this surface is managed through the provided playbooks.
Audit Metadata