maintain-verification-skill
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the project's source code and feature map files, creating a vulnerability surface for indirect instructions. \n
- Ingestion points: SKILL.md, feature files in the
features/directory, and project source code read by subagents in Step 2. \n - Boundary markers: There are no instructions to use specific delimiters or warnings to isolate ingested data from the agent's control logic. \n
- Capability inventory: The skill can execute local harness scripts, modify files in its local skill directory, and generate pull requests. \n
- Sanitization: The skill lacks explicit sanitization or escaping mechanisms for content extracted from source files.\n- [SAFE]: The skill operates within a restricted scope, focusing on a project-local directory and explicitly prohibiting changes to the product's functional code. It incorporates robust operational checks, including process health monitoring and automated cleanup of residues after execution iterations.
Audit Metadata