PyTorch Documentation Search MCP server

This MCP server provides semantic search capabilities over PyTorch documentation, allowing users to find relevant documentation, APIs, code examples, and error messages using vector embeddings and semantic similarity for high-quality search results.

Installation

Environment Setup

Create a conda environment with all dependencies:

conda env create -f environment.yml
conda activate pytorch_docs_search

For a minimal environment:

conda env create -f minimal_env.yml
conda activate pytorch_docs_search_min

API Key Setup

The tool requires an OpenAI API key for generating embeddings:

export OPENAI_API_KEY=your_key_here

Usage

MCP Integration Options

You can integrate the tool with Claude Code in three ways:

1. Direct STDIO Integration (Local Development)

# Register with Claude CLI
./register_mcp.sh

This registers the tool with Claude using the STDIO transport.

2. SSE Integration (Server Deployment)

# Start the server
python -m ptsearch.server --transport sse --host 0.0.0.0 --port 5000

# Register with Claude CLI
claude mcp add search_pytorch_docs http://localhost:5000/events --transport sse

3. UVX Integration (Packaged Distribution)

# Run with UVX
./run_mcp_uvx.sh

This starts the server using UVX transport.

Using with Claude Code

Once registered with Claude Code, you can ask questions about PyTorch, and Claude will automatically use the search tool:

How do I implement a custom dataset in PyTorch?

Direct CLI Usage

You can also use the tool directly from the command line:

# Search from command line
python -m ptsearch.server --transport stdio --data-dir ./data

Search Features

The tool provides:

• Semantic search for PyTorch documentation
• Code-aware search results (differentiates between code and text)
• Configurable search parameters and result formatting

How to add this MCP server to Cursor

There are two ways to add an MCP server to Cursor. The most common way is to add the server globally in the ~/.cursor/mcp.json file so that it is available in all of your projects.

If you only need the server in a single project, you can add it to the project instead by creating or adding it to the .cursor/mcp.json file.

Adding an MCP server to Cursor globally

To add a global MCP server go to Cursor Settings > MCP and click "Add new global MCP server".

When you click that button the ~/.cursor/mcp.json file will be opened and you can add your server like this:

{
    "mcpServers": {
        "cursor-rules-mcp": {
            "command": "npx",
            "args": [
                "-y",
                "cursor-rules-mcp"
            ]
        }
    }
}

Adding an MCP server to a project

To add an MCP server to a project you can create a new .cursor/mcp.json file or add it to the existing one. This will look exactly the same as the global MCP server example above.

How to use the MCP server

Once the server is installed, you might need to head back to Settings > MCP and click the refresh button.

The Cursor agent will then be able to see the available tools the added MCP server has available and will call them when it needs to.

You can also explictly ask the agent to use the tool by mentioning the tool name and describing what the function does.