Skills
skills/togethercomputer/skills/together-embeddings/Gen Agent Trust Hub

together-embeddings

Pass

Audited by Gen Agent Trust Hub on Mar 31, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The retrieval-augmented generation (RAG) implementation in scripts/rag_pipeline.py exhibits an indirect prompt injection surface common to such architectures.
  • Ingestion points: Contextual data from the knowledge list and user queries in scripts/rag_pipeline.py.
  • Boundary markers: Uses structured system messages to isolate context from queries.
  • Capability inventory: Interacts with the client.chat.completions.create endpoint.
  • Sanitization: No specific content sanitization is applied prior to prompt interpolation.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 31, 2026, 07:24 PM
Security Audit — agent-trust-hub — together-embeddings