lmql-constraints
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security issues detected. The skill is a documentation resource for the LMQL library.
- [INDIRECT_PROMPT_INJECTION]: The skill describes patterns for processing untrusted data via LLM prompts. 1. Ingestion points: File 'skill.md' (functions 'greet', 'constrained_summary', 'typed_sentiment', 'async_classify'). 2. Boundary markers: None present. 3. Capability inventory: LLM generation via the 'lmql' library. No high-privilege operations. 4. Sanitization: None demonstrated.
- [EXTERNAL_DOWNLOADS]: References the installation of the 'lmql' library via pip.
- [CREDENTIALS_UNSAFE]: Uses placeholders ('...') for API keys in configuration examples, which is a safe practice.
Audit Metadata