pinecone-research

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill ingests external data from local files and command-line arguments to populate a vector database for agent retrieval.
  • Ingestion points: scripts/rag_pipeline.py reads content from all .txt files within a user-provided directory. scripts/memory_manager.py accepts arbitrary strings via the --text CLI argument.
  • Boundary markers: The instructions do not define delimiters or specific agent guidance to treat retrieved context as untrusted or separate from system instructions.
  • Capability inventory: The skill performs network operations to external API endpoints for Pinecone and OpenAI services to manage vector embeddings and search.
  • Sanitization: No filtering or validation is applied to the ingested text before it is stored or retrieved.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 03:31 AM
Security Audit — agent-trust-hub — pinecone-research