nonatofabio/local-faiss-mcp
🔌 MCP Servernonatofabio
A local FAISS-powered MCP server for efficient RAG, document ingestion, and semantic search capabilities.
local-faiss-mcp is a specialized Model Context Protocol (MCP) server designed to bring high-performance vector search to local AI environments. By leveraging the FAISS (Facebook AI Similarity Search) library, it allows users to perform fast, scalable similarity searches on their own hardware. The server handles the entire pipeline of document processing, including parsing common file formats like PDF, Markdown, TXT, and DOCX.
Key features include built-in support for semantic search and re-ranking, which significantly improves the relevance of retrieved context in RAG applications. The project also includes a set of CLI tools that simplify the indexing of local documents and the execution of queries, allowing for seamless integration with MCP-compliant clients. This tool is ideal for developers building privacy-focused AI agents that require efficient, local-first retrieval augmented generation without relying on external cloud-based vector database services.
💡Highlights
- ├─Supports PDF, TXT, MD, and DOCX
- ├─Local FAISS vector indexing
- └─Built-in semantic re-ranking
🎯For
- ├─AI Engineers
- └─RAG Developers