simplify
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFENO_CODEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill content consists exclusively of Markdown documentation and metadata. No malicious patterns, such as prompt injection or unauthorized data access, were detected.- [NO_CODE]: The skill includes no executable scripts, binaries, or configuration files that would perform operations on the host system. It defines an operational contract for the agent to follow.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted code during refactoring tasks.
- Ingestion points: Source code files provided to the agent for simplification (SKILL.md).
- Boundary markers: The 'Understand Before Touching' phase requires the agent to identify code responsibilities and edge cases before modification.
- Capability inventory: No scripts or tool execution capabilities are defined within the skill files (codemap.md).
- Sanitization: Verification relies on a checklist requiring that existing tests and build processes pass.
Audit Metadata