together-chat-completions
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: As a chat interface, the skill processes external data (user messages and tool outputs) which is the primary vector for indirect prompt injection. The implementation mitigates this by using the official SDK's message structure and recommending JSON schemas for tool calling and structured outputs, which aids in maintaining context boundaries.
- [SAFE]: The skill uses official vendor libraries and points to legitimate API endpoints. Credentials are handled via environment variables, and no instances of command injection, persistence, or data exfiltration were found across the scripts and documentation.
Audit Metadata