langchain-architecture

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.80). Yes — SKILL.md explicitly shows loading a web search tool (load_tools(["serpapi", ...]) in the Quick Start) and RAG/document-loader patterns (RetrievalQA, document_loaders, DirectoryLoader) that ingest open/public or user-provided content which the agent reads and can use to choose actions, exposing it to untrusted third-party content.