rag-implementation

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill implements Retrieval-Augmented Generation (RAG) patterns which ingest external data into the agent's context, creating a vulnerability surface for indirect prompt injection.
  • Ingestion points: The implementation uses PineconeVectorStore, WeaviateVectorStore, and other retrievers to fetch external document content into the prompt context (SKILL.md, references/details.md).
  • Boundary markers: Prompt templates use basic text headers (e.g., "Context:") but lack robust delimiters or specific instructions to the model to ignore instructions embedded within the retrieved data (SKILL.md, references/details.md).
  • Capability inventory: The provided code examples focus on information retrieval and natural language generation; they do not include high-risk capabilities such as arbitrary shell execution or system-level file writes.
  • Sanitization: The implementation patterns do not include logic for sanitizing or filtering instructions from retrieved document content before interpolation.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 11:55 AM
Security Audit — agent-trust-hub — rag-implementation