Xiaoyao Search

v1.0.0Search & Data Extractionstable

小遥搜索,听懂你的话、看懂你的图,用AI找到本地任何文件。让搜索像聊天一样简单。XiaoyaoSearch: Understands your words, reads your images, finds any local file with AI. Making search as easy as chatting.

agent-skillsai-searchdocument-searchfile-searchlocal-search
Share:
1,021
Stars
0
Downloads
0
Weekly
0/5

What is Xiaoyao Search?

Xiaoyao Search is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to 小遥搜索,听懂你的话、看懂你的图,用ai找到本地任何文件。让搜索像聊天一样简单。xiaoyaosearch: understands your words, reads your images, finds any local file with ai. making search as easy as chatting.

小遥搜索,听懂你的话、看懂你的图,用AI找到本地任何文件。让搜索像聊天一样简单。XiaoyaoSearch: Understands your words, reads your images, finds any local file with AI. Making search as easy as chatting.

This server falls under the Search & Data Extraction category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • 小遥搜索,听懂你的话、看懂你的图,用AI找到本地任何文件。让搜索像聊天一样简单。XiaoyaoSearch: Under

Use Cases

Natural language local file search
Multimodal AI-powered document discovery
dtsola

Maintainer

LicenseNOASSERTION
Languagepython
Versionv1.0.0
UpdatedMay 22, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx xiaoyaosearch

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Xiaoyao Search

XiaoYao Search is a locally-hosted AI-powered file search engine that understands natural language queries and can even analyze images to find matching files on your computer. It combines semantic search (via BGE-M3 embeddings), full-text keyword search (via Whoosh), voice input (via FasterWhisper), and visual similarity matching (via CN-CLIP) into a single hybrid search system. Developers and power users connect to it through an MCP SSE endpoint, allowing Claude or other AI assistants to invoke local file searches as tool calls.

Prerequisites

  • Python 3.10+ with pip for the backend service
  • Node.js 18+ for the frontend (optional UI)
  • An MCP client such as Claude Desktop that supports SSE transport
  • Sufficient local disk space to store FAISS vector indexes and Whoosh full-text indexes
  • Optional: microphone access and GPU for faster embeddings
1

Clone the repository

Clone the XiaoYao Search repository from GitHub to get both the backend and frontend source code.

git clone https://github.com/dtsola/xiaoyaosearch.git
cd xiaoyaosearch
2

Install Python backend dependencies

Install the required Python packages including vector search libraries, the Whoosh indexer, and FasterWhisper for voice input support.

cd backend
pip install -r requirements.txt
pip install faster-whisper
3

Configure environment variables

Create a .env file in the backend directory to set paths for the FAISS vector index, Whoosh full-text index, SQLite database, and to enable MCP SSE support.

FAISS_INDEX_PATH=../data/indexes/faiss
WHOOSH_INDEX_PATH=../data/indexes/whoosh
DATABASE_PATH=../data/database/xiaoyao_search.db
API_HOST=127.0.0.1
API_PORT=8000
MCP_SSE_ENABLED=true
4

Start the backend server

Launch the Python backend which starts the MCP SSE endpoint at http://127.0.0.1:8000/mcp. The first run will build search indexes from your files.

python main.py
5

Configure your MCP client

Add XiaoYao Search to your Claude Desktop (or other MCP client) configuration using the SSE transport type pointing to the running backend.

{
  "mcpServers": {
    "xiaoyaosearch": {
      "type": "sse",
      "url": "http://127.0.0.1:8000/mcp"
    }
  }
}

Xiaoyao Search Examples

Client configuration

SSE-based MCP client configuration for Claude Desktop pointing to the local XiaoYao Search backend.

{
  "mcpServers": {
    "xiaoyaosearch": {
      "type": "sse",
      "url": "http://127.0.0.1:8000/mcp"
    }
  }
}

Prompts to try

Example natural language queries you can send to Claude once XiaoYao Search is connected.

- "Find documents about async programming on my computer"
- "Search for files containing the keyword 'budget 2024'"
- "Find images that look similar to my landscape photo"
- "Locate any PDF files I created last month related to the project proposal"

Troubleshooting Xiaoyao Search

MCP client cannot connect to the SSE endpoint

Ensure MCP_SSE_ENABLED=true is set in your .env file and the backend is running. Verify the backend is listening on port 8000 with: curl http://127.0.0.1:8000/mcp

Search returns no results after startup

The FAISS and Whoosh indexes must be built before searching. Trigger an index rebuild from the UI or by running the indexing script manually. Check that FAISS_INDEX_PATH and WHOOSH_INDEX_PATH point to writable directories.

Voice search fails or produces poor transcriptions

Ensure faster-whisper is installed and your microphone is accessible. Try changing the whisper model size in config (e.g. 'small' for speed vs 'medium' for accuracy). On macOS you may need to grant microphone permissions to the terminal.

Frequently Asked Questions about Xiaoyao Search

What is Xiaoyao Search?

Xiaoyao Search is a Model Context Protocol (MCP) server that 小遥搜索,听懂你的话、看懂你的图,用ai找到本地任何文件。让搜索像聊天一样简单。xiaoyaosearch: understands your words, reads your images, finds any local file with ai. making search as easy as chatting. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Xiaoyao Search?

Follow the installation instructions on the Xiaoyao Search GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Xiaoyao Search?

Xiaoyao Search works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Xiaoyao Search free to use?

Yes, Xiaoyao Search is open source and available under the NOASSERTION license. You can use it freely in both personal and commercial projects.

Browse More Search & Data Extraction MCP Servers

Explore all search & data extraction servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "xiaoyaosearch": { "command": "npx", "args": ["-y", "xiaoyaosearch"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Xiaoyao Search?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides