WindTools MCP Server

Name: WindTools MCP Server
Author: ZahidGalea

@ZahidGalea

Repository

About WindTools MCP Server

Your own codebase tools like code semantic search

#python #mcp #developer-tools #mcp-server

Basic information

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "windtools-mcp": {
      "command": "uv",
      "args": [
        "sync",
        "--dev"
      ]
    }
  }
}

Tools

`directory_path` (string): Path to list contents of, should be absolute path to a directory

JSON string containing directory information including file types and sizes

JSON string with initialization status of ChromaDB and embedding model

`target_directories` (array of strings): List of absolute paths to directories to index

JSON string containing indexing statistics and results

`query` (string): Search query describing what you're looking for

JSON string containing search results with relevant code snippets

Overview

What is WindTools MCP Server?

WindTools MCP Server is an MCP (Model Context Protocol) server for code assistants, providing document embedding and retrieval capabilities using ChromaDB and sentence transformers. It indexes code repositories and enables semantic search over code snippets, designed for developers who need to integrate vector-based code search into AI‑powered tooling.

How to use WindTools MCP Server?

Install the package via pip install windtools-mcp and configure it with environment variables such as DATA_ROOT and SENTENCE_TRANSFORMER_PATH. Add it to your MCP client (e.g., Claude Desktop) by specifying the command uvx -p 3.11 windtools-mcp in the client’s configuration JSON.

Key features of WindTools MCP Server

Semantic code search using sentence transformers
Automatic indexing of code files from specified directories
Persistent embedding storage in ChromaDB
Built‑in tools for directory exploration
Background initialization for minimal startup delay
Environment‑variable configuration for paths and models

Use cases of WindTools MCP Server

Search for relevant code snippets across large repositories
Index multiple directories for unified code retrieval
Explore and navigate codebases via an MCP client
Integrate semantic code search into AI code assistants

FAQ from WindTools MCP Server

What are the runtime requirements?

WindTools MCP Server requires Python 3.11, as ChromaDB has compatibility issues with newer Python versions.

Where is the data stored?

Data (ChromaDB database and model cache) is stored in the directory specified by the DATA_ROOT environment variable, defaulting to a data folder inside the package.

How does the initialization process work?

The server initializes ChromaDB and the embedding model asynchronously in the background, allowing it to accept requests immediately. Use the get_initialization_status tool to check when initialization is complete.

What embedding model is used?

The default sentence transformer model is jinaai/jina-embeddings-v2-base-code, configurable via the SENTENCE_TRANSFORMER_PATH environment variable.