This is a RAG knowledge base server that handles semantic search, document chunking, and relevance ranking to reduce hallucination in AI responses. The source shows it's MCP protocol compliant and installable via pip or Smithery, but doesn't detail specific tools or prompts exposed. You'd reach for this when you need to ground Claude's responses in your own documents with proper retrieval mechanics. Built by MEOK AI Labs with a focus on EU AI Act compliance including risk management and bias detection requirements. The README emphasizes enterprise features and compliance tooling more than the underlying RAG implementation details, so expect a batteries-included approach to document retrieval rather than granular control over embedding models or chunking strategies.
claude mcp add --transport stdio csoai-org-rag-knowledge-mcp -- uvx rag-knowledge-mcp