MCP Code Snippets

This is an local adapter for MCP server that allows Agents to lookup code snippets in your project. Local MCP implementation allows to auto-fill some arguments for the remote MCP server, and therefore make lookup more accurate. If no results are found with these arguments, results without arguments are returned. By default, it uses Qdrant-maintaned remote MCP server with a collection of code snippets, but you can use your own collection by providing a custom MCP proxy configuration.

Setup

mcp-code-snippets should be available as an executable in your system, so that MCP client (like Cursor) can find it.

Global installation

Simplest option is to install it globally with pip:

pip install mcp-code-snippets

In this way it will be available in your system path, and can be used by any MCP client.

Project installation

Install mcp-code-snippets with a package manager of your choice in your project.

An example using uv:

uv add mcp-code-snippets

Note: it might be worth it to install mcp-code-snippets as a dev dependency, as it is not needed in production.

Usage

To start using mcp-code-snippets in your project, you need to configure MCP client to use it.

Cursor

Add the following to your .cursor/mcp.json file:

{
    "mcpServers": {
        "mcp-code-snippets": {
            "command": "mcp-code-snippets"
        }
    }
}

Advanced configuration

mcp-code-snippest has one cli argument: --transport, which can be set either to stdio, sse or streamable-http. The default is stdio. There is not much other configuration available at the moment, though more to be added.

Other than cli arguments, there are a couple of environment variables to configure the tool: PROJECT_ROOT_PATH and MCP_PROXY_CONFIG.

PROJECT_ROOT_PATH is the path to the project root. It is used to detect the programming language of the project, and to extract dependencies. There is an automatic detection of the project language, but you can also specify it manually by setting PROGRAMMING_LANGUAGE environment variable.

For Python supported package managers are those following PyPA specification, poetry and uv syntax beyond PyPA is also supported.

MCP_PROXY_CONFIG is the path to the MCP proxy configuration file, which is used to configure the access to the MCP server with the code snippets.

An example would be:

{
     "mcpServers": {
         "mcp-server-qdrant": {
             "url": "https://mcp.qdrant.tech/mcp/",
             "transport": "streamable-http"
         }
     }
}

Editors integrations

You can use mcp-code-snippets with various editors that support MCP protocol. An example configuration for Cursor in .cursor/mcp.json would look like this:

{
    "mcpServers": {
        "mcp-code-snippets": {
            "command": "mcp-code-snippets",
            "env": {
                "MCP_PROXY_CONFIG": "/home/user/my-mcp/mcp_proxy_example.json",
                "PROJECT_ROOT_PATH": "/home/user/my-project"
            }
        }
    }
}

MCP_PROXY_CONFIG default value is mcp_proxy.json, and PROJECT_ROOT_PATH defaults to the current directory.

NOTE: Cursor might not connect to the MCP server with an error like ENOENT mcp-code-snippets, in this case you might need to use absolute paths for all envs and commands.

Development

This project uses uv for package management and ruff for linting and formatting. In order to build the package just run:

uv build

After building, try running the package with:

mcp-code-snippets --help