knowledge-rag

Overview

Knowledge RAG is a comprehensive local-first RAG system that enables instant search across various document formats without requiring any external servers or API keys. It implements hybrid search combining BM25 keyword search, semantic vectors, and cross-encoder reranking for high precision results. The system is built with Python using FastMCP framework and provides 12 MCP tools for full CRUD operations, search, evaluation, and document management.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• searchquery: str

  Search through indexed documents using hybrid search with semantic and keyword matching

  Try:Search for documents about machine learningTry:Find files related to cybersecurity threats
• getdoc_id: str

  Retrieve a specific document by its ID

  Try:Get document with ID abc123Try:Retrieve the content of doc xyz789
• addpath: str

  Add a new document to the index

  Try:Add the report.pdf to the knowledge baseTry:Index the meeting_notes.md file
• updatepath: str

  Update an existing document in the index

  Try:Update the project_requirements.docxTry:Refresh the technical_specs.txt
• removepath: str

  Remove a document from the index

  Try:Remove the outdated_policy.pdfTry:Delete the deprecated_code.py file
• reindex

  Rebuild the entire index from scratch

  Try:Rebuild the entire knowledge baseTry:Refresh all indexed documents
• list

  List all documents in the index

  Try:Show all indexed documentsTry:List all files in the knowledge base
• stats

  Get statistics about the index

  Try:Show index statisticsTry:How many documents are indexed?
• urlurl: str

  Index content from a web URL

  Try:Index this article https://example.com/securityTry:Add documentation from https://api.example.com/docs
• similardoc_id: str

  Find documents similar to a reference document

  Try:Find documents similar to the security audit reportTry:Show me files like the project plan
• evaluate

  Evaluate retrieval performance with MRR@5 and Recall@5 metrics

  Try:Evaluate search performanceTry:Check retrieval quality metrics

Comparable tools

Installation

pip install knowledge-rag

{
  "mcpServers": {
    "knowledge-rag": {
      "command": "python",
      "args": ["-m", "knowledge_rag.server"],
      "env": {}
    }
  }
}

FAQ

What file formats are supported?

Knowledge RAG supports 20+ formats including PDF, Markdown, Python code, Word documents, Excel files, Jupyter notebooks, and various programming languages (C++, JavaScript, TypeScript, etc.).

Is it truly local with no cloud dependencies?

Yes, all processing runs locally via ONNX. No servers, API keys, or data leaves your machine.

On Hacker News

Recent discussion from the developer community.

• Knowledge-RAG – Local RAG for Claude Code with hybrid search and cross-encoder▲ 2
  Story by lyonzin · 2026-03-19

Compare knowledge-rag with