A practical POC demonstrating how to build and run a local MCP server with Retrieval-Augmented Generation (RAG) for semantic search over internal documentation. Leverages Node.js, TypeScript, Hugging Face embeddings, and an in-memory vector store to enable fast, context-aware answers in tools like Cursor.
mcp semantic-search rag huggingface langchain retrieval-augmented-generation model-context-protocol mcp-server mcp-rag mcp-server-setup mcp-rag-server
-
Updated
May 13, 2025 - TypeScript