loop-me
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions define an interactive process for gathering workflow requirements and saving them to
workflows/*.mdandNOTES.md. This behavior aligns with the stated purpose and lacks any malicious patterns or unauthorized operations. - [INDIRECT_PROMPT_INJECTION]: The skill reads user-controlled content to generate specifications, which is a standard attack surface for agents processing external data. Although present, the risk is negligible as it is restricted to local file operations.
- Ingestion points:
NOTES.mdand interactive user chat. - Boundary markers: None specified in the instructions.
- Capability inventory: Creation, modification, and deletion of markdown files in the
workflows/directory. - Sanitization: None specified in the instructions.
Audit Metadata