langchain-rag

Pass

Audited by Gen Agent Trust Hub on Aug 16, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill provides implementation templates for Retrieval Augmented Generation (RAG) that ingest data from external sources.
  • Ingestion points: The code examples utilize CheerioWebBaseLoader for web scraping and PDFLoader for local file ingestion (SKILL.md).
  • Boundary markers: The system prompt template (使用以下上下文回答问题:\n\n${context}) lacks explicit delimiters (e.g., XML tags) or instructions to the model to ignore any instructions embedded within the retrieved context.
  • Capability inventory: The skill's templates do not include high-risk capabilities such as arbitrary command execution, file system modifications, or network exfiltration logic beyond the loaders themselves.
  • Sanitization: There is no demonstrated sanitization or filtering of the retrieved content before it is interpolated into the model prompt.
  • [EXTERNAL_DOWNLOADS]: The skill includes code examples that fetch content from https://docs.langchain.com. This targets the official documentation of a well-known service and is a standard developmental use case.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 16, 2026, 12:30 AM
Security Audit — agent-trust-hub — langchain-rag