langchain-neo4j
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates Graph Q&A and vector-based retrieval which inherently processes data from an external Neo4j database. This creates a surface for indirect prompt injection if the database content is attacker-controlled. The skill provides clear documentation on the necessary security flags and configuration to mitigate this risk.
- Ingestion points: Neo4j database content accessed via Neo4jGraph, Neo4jVector, and GraphCypherQAChain.
- Boundary markers: The documentation explicitly highlights the requirement for allow_dangerous_requests=True in the GraphCypherQAChain as a security acknowledgment.
- Capability inventory: The skill performs database read and write operations via Cypher queries generated from natural language.
- Sanitization: The Neo4jGraph class supports a sanitize parameter to filter large list properties from database results, reducing potential exposure of sensitive or excessive data.
- [EXTERNAL_DOWNLOADS]: The skill correctly identifies its dependencies, such as the official langchain-neo4j package and the standard Neo4j Docker image. These resources originate from well-known and trusted organizations, posing no security risk in this context.
Audit Metadata