outlines
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
- [SAFE]: The skill provides legitimate documentation and code examples for the 'outlines' library, focusing on structured output generation and Pydantic integration.
- [SAFE]: The skill promotes security best practices by illustrating how to enforce schemas, numeric constraints, and regex patterns on LLM outputs to prevent malformed data.
- [INDIRECT_PROMPT_INJECTION]: The skill includes patterns for data extraction from external sources (e.g., resumes, invoices). This presents an indirect prompt injection surface common to data processing tasks, which the skill mitigates through rigorous Pydantic schema validation.
- [EXTERNAL_DOWNLOADS]: Installation instructions for standard Python packages (outlines, vllm, transformers) and references to reputable model repositories (Hugging Face) are provided.
- [DYNAMIC_EXECUTION]: The skill demonstrates how to generate Python and SQL code structures based on user input, though it does not provide mechanisms for the automated execution of the generated content.
Audit Metadata