together-cost-tuning
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches live pricing information and model eligibility data from the official Together AI documentation (docs.together.ai) and references the official SDK and skills repositories under the togethercomputer GitHub organization.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data from usage logs and billing analytics which presents a potential surface for indirect prompt injection.
- Ingestion points: The
Read,Glob, andGreptools are used to ingest usage window data, request counts, and billing analytics as specified in SKILL.md. - Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are defined for the ingested usage data.
- Capability inventory: Includes
WebFetchfor pricing data andWrite/Editfor updating cost models and instrumentation logic. - Sanitization: No explicit sanitization or filtering of the external usage data is mentioned in the instructions.
Audit Metadata