MCP Local Dev

MCP Local Dev lets you delegate the setup and management of local development environments to AI, handling environment configuration, dependencies, and testing so you can focus on building features. It provides isolated, sandboxed runtimes that automatically pick the right package manager and keep environments clean and easy to reproduce across projects.

How to use

You interact with MCP Local Dev through an MCP client. Start by ensuring the local MCP server is configured, then connect the client to the server and request a development environment for a specific project. The system will detect your runtime (Python, Node.js, or Bun), pick the best package manager, and spin up a sandboxed environment. You can run tests, check coverage, and iterate with the AI assistant to explore, modify, or clean up the environment as needed.

How to install

Prerequisites: you should have a working MCP client and a runtime environment to run the MCP server. Install or enable the client according to your setup, then prepare the local server command configured underneath.

Add the MCP server configuration to your client so it can launch the local dev environment. Use the exact command and arguments shown below to ensure proper initialization.

{
  "servers": {
     "local_dev": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/txbm/mcp-local-dev@main",
        "mcp-local-dev"
      ]
    }
  }
}

Core concepts you’ll use

• Automatic runtime detection for Python, Node.js, and Bun
• Smart package manager selection for each runtime
• Sandboxed environments for isolation
• Automatic cleanup of unused environments
• GitHub repository support for projects
• Local project support for quick, repeatable setups

Environment and runtime details

Environments are isolated per project and connect to your installed runtimes. The system uses the fastest available package manager for each runtime and allows full network access for package operations. Runtimes are managed without requiring custom configuration for each project, and environments are cleaned up automatically when you’re done.