clean-up
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data in the form of code diffs and working-tree states, which could potentially contain malicious instructions designed to influence agent behavior.
- Ingestion points: The skill reads the "recent diff and current working-tree state" as specified in
SKILL.md. - Boundary markers: The instructions include scope limitations such as "Limit cleanup to the requested change" and "Preserve unrelated edits," but lack explicit technical delimiters for external content.
- Capability inventory: The skill is authorized to perform "scoped edits" (writing to files) and "rerun affected checks" (executing shell commands).
- Sanitization: There are no instructions for sanitizing or escaping the content of the code diffs before they are processed by the agent.
- [COMMAND_EXECUTION]: The skill instructions authorize the agent to "Verify in proportion to the change using relevant existing checks" and "rerun affected checks," which involves executing local testing and linting tools via the shell.
Audit Metadata