code-polish
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external code as input and generates modifications to those files.
- Ingestion points: The skill reads entire modules or files provided by the user (Phase 0 in
SKILL.md). - Boundary markers: The instructions do not specify the use of delimiters or clear separation between the agent's instructions and the untrusted source code content.
- Capability inventory: The skill is designed to rewrite files and modify content (specifically comments and formatting).
- Sanitization: The 'Prime Directive' and Phase 4 (Verification) provide significant mitigations by strictly prohibiting logic changes and requiring behavioral identity checks, reducing the risk of an injection successfully altering the program's execution flow.
Audit Metadata