integrating-ai-web-apps

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly ingests and uses untrusted public content (e.g., LangChain's WikipediaQueryRun in references/LANGCHAIN-JS.md, the MCP tool "fetch-data" that fetches https://api.example.com/search in references/MCP-INTEGRATION.md, and RAG/document ingestion workflows in RAG-AND-SUMMARIZATION.md) as part of agent workflows where retrieved third‑party content is used to drive tool calls and responses, enabling indirect prompt injection.