llm-application-dev-langchain-agent
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides implementation patterns for RAG pipelines and tool-calling agents that ingest and process data from external vector databases and user-supplied queries, which could contain malicious instructions.
- Ingestion points: Untrusted data enters the agent's context through retrieved documents in
PineconeVectorStoreand user queries passed totool_functioninSKILL.md. - Boundary markers: The skill includes instructions in the "Limitations" section to validate tool arguments and user authority, though specific delimiters for external content are not shown in the templates.
- Capability inventory: The skill facilitates the creation of tools that perform
external_calloperations and provides patterns for a FastAPI server to invoke agents, potentially executing actions based on processed data. - Sanitization: The skill advises returning redacted tool errors and using environment variables for secret management to mitigate data exposure risks.
Audit Metadata