together-embeddings
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill correctly directs the user to use the official Together AI libraries (together>=2.0.0 for Python and together-ai for Node.js) and makes authorized API calls to the official Together AI endpoint at api.together.xyz.
- [INDIRECT_PROMPT_INJECTION]: The scripts/rag_pipeline.py file implements a standard RAG architecture.
- Ingestion points: Untrusted data from a knowledge base is retrieved and inserted into the system prompt in the rag_query function.
- Boundary markers: The retrieved context is preceded by a 'Context:' header.
- Capability inventory: The skill uses the Together AI SDK to generate embeddings and retrieve text completions.
- Sanitization: The skill does not perform specific sanitization on the retrieved text, which is typical for a RAG reference implementation.
- Note: The implementation is consistent with the skill's documented purpose as a RAG developer guide and does not exhibit malicious behavior.
Audit Metadata