rmslop
Pass
Audited by Gen Agent Trust Hub on Nov 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted code changes (uncommitted changes or branch diffs). Malicious instructions could be embedded within the code or comments being cleaned, which might attempt to influence the agent's behavior during the editing process.
- Ingestion points: The skill ingests code content from git diffs and local files (SKILL.md).
- Boundary markers: No explicit delimiters or instructions are provided to the agent to treat the code content strictly as data rather than instructions.
- Capability inventory: The agent likely uses tools to read file content, list git changes, and apply modifications to the filesystem.
- Sanitization: No sanitization, validation, or filtering of the ingested code content is specified before processing.
Audit Metadata