agent-md-refactor
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes content from existing documentation files like CLAUDE.md or AGENTS.md, which could potentially contain untrusted instructions. However, the skill provides specific workflows to identify contradictions and prune redundant or non-actionable content, which serves as a basic verification step. * Ingestion points: Input documentation files provided by the user for refactoring. * Capability inventory: Documentation analysis and markdown file creation/linking. * Sanitization: Includes a manual verification step for contradictions and automated pruning of vague instructions.
- [SAFE]: No malicious patterns, network exfiltration, or dangerous command executions were detected. The skill strictly operates on markdown documentation within the project environment.
Audit Metadata