together-evaluations
Pass
Audited by Gen Agent Trust Hub on Mar 4, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill incorporates official client libraries from Together AI, specifically the 'together' Python package and 'together-ai' Node.js package, sourced from official registries.
- [SAFE]: All identified network communications are directed to Together AI's legitimate API endpoints (api.together.xyz) and supported external providers (OpenAI, Anthropic, Google).
- [SAFE]: No hardcoded secrets were detected; the skill provides clear guidance on using environment variables for managing API tokens.
- [SAFE]: While the skill ingests model outputs for evaluation, this is the primary and intended function of the tool. The implementation uses standard SDK methods and does not bypass any agent safety filters.
Audit Metadata