mcp-local-rag

概述

mcp-local-rag是一个注重隐私的本地化RAG类网页搜索MCP服务器，完全无需外部API依赖。它支持多个搜索引擎，包括DuckDuckGo、Google、Bing、Brave、Wikipedia等，并通过语义相似性排序提供相关结果。该服务器已容器化，可与Claude Desktop、Cursor和Goose等多种MCP客户端配合使用。它还包含Agent技能，教导LLM进行有效的网页搜索和深度研究。

试试问 AI

装完之后，这里有 3 个你可以让 AI 做的事：

• deep_research

  Comprehensive multi-engine research across multiple search backends

  试试:Do deep research on recent quantum computing developments试试:Search multiple sources for sustainable energy solutions
• deep_research_google

  Google-focused deep dive with technical/scientific queries

  试试:Find comprehensive technical documentation about Kubernetes optimization试试:Research the latest advancements in machine learning models
• deep_research_ddgs

  Privacy-first deep research using DuckDuckGo

  试试:Research general topics without tracking试试:Find information about health conditions privately
• rag_search_ddgs

  Quick single search using DuckDuckGo for fast answers

  试试:What's the weather forecast for tomorrow in London?试试:Who won the latest Oscar for Best Picture?
• rag_search_google

  Quick single search using Google for focused results

  试试:Find recent news about artificial intelligence breakthroughs试试:Search for documentation on Python async programming

可对比工具

安装

{
  "mcpServers": {
    "mcp-local-rag": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "--init",
        "-e",
        "DOCKER_CONTAINER=true",
        "ghcr.io/nkapila6/mcp-local-rag:v1.0.2"
      ]
    }
  }
}

{
  "mcpServers": {
    "mcp-local-rag":{
      "command": "uvx",
      "args": [
        "--python=3.10",
        "--from",
        "git+https://github.com/nkapila6/mcp-local-rag",
        "mcp-local-rag"
      ]
    }
  }
}

mcp-local-rag 对比