together-rate-limits
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of dynamic rate-limit data from Together AI API response headers.
- Ingestion points: The skill instructions specify capturing x-ratelimit-* and x-tokenlimit-* fields from serverless inference responses in SKILL.md.
- Boundary markers: The instructions recommend using redacted responses to capture these fields, acting as a manual boundary to prevent processing full, potentially malicious, payloads.
- Capability inventory: The skill utilizes Read, Glob, Grep, WebFetch, Write, and Edit tools to analyze current configurations and implement limiters.
- Sanitization: No programmatic sanitization or validation of the header values is explicitly defined, relying on the agent's interpretation of numeric data.
- [EXTERNAL_DOWNLOADS]: The skill references the official togethercomputer GitHub organization for Python SDKs and agent skills, as well as documentation at docs.together.ai. These are recognized as legitimate, well-known resources for the Together AI service.
Audit Metadata