1

You have a single MCP server that can automatically discover, configure, and manage multiple MCP servers on your machine, whether they’re remote or local. This means you install one MCP server, and it handles the rest of the setup and coordination for you.

How to use

You can start by choosing a remote setup for ease of use, or run locally using prebuilt options like Docker or npm, or run directly from the source. The MCP server supports both quick, remote configuration and deeper, local experimentation.

How to install

Prerequisites depend on how you want to run the MCP server. You can use a remote hosted endpoint for simplicity, run a local Docker container, use npx, or run from source.

Remote setup minimal steps

• Add the remote MCP configuration to your client as shown below to enable remote discovery and management.

Local setup options
- Docker: run the container locally and expose the port you need.
- npx: execute the MCP server package directly from npm.
- From source: clone the repository and run the development server with the provided tooling.

Configuration and examples

You can configure MCP endpoints in JSON. The following examples show how to point your client to a remote MCP server. You can also add a filesystem server to give your LLM file-system access if needed.

{
  "mcpServers": {
    "1mcpserver": {
      "url": "https://mcp.1mcpserver.com/mcp/",
      "headers": {
        "Accept": "text/event-stream",
        "Cache-Control": "no-cache"
      }
    }
  }
}

HTTP and local runtime options

HTTP remote server configuration is shown below. This is the recommended remote setup for simplicity.

{
  "mcpServers": {
    "1mcpserver": {
      "url": "https://mcp.1mcpserver.com/mcp/",
      "headers": {
        "Accept": "text/event-stream",
        "Cache-Control": "no-cache"
      }
    }
  }
}

If you want to run locally using Docker, use the following command to start the MCP server locally.

docker run -p 8080:8080 ghcr.io/particlefuture/1mcpserver:latest

If you prefer running with stdio mode instead of streamable HTTP, you can run the container with stdio input/output.

run --rm -i ghcr.io/particlefuture/1mcpserver:latest --local

A minimal local configuration for the MCP server when using remote setup is shown here.

{
  "mcpServers": {
    "1mcpserver": {
      "url": "https://mcp.1mcpserver.com/mcp/"
    }
  }
}

Local from source setup

If you want to run the MCP server directly from the source code, follow these steps.

git clone https://github.com/particlefuture/MCPDiscovery.git
cd MCPDiscovery
uv sync
uv run server.py --local

Configuration to run the locally sourced server can point to the cloned directory and run the local server as shown.

{
  "mcpServers": {
    "1mcpserver": {
      "command": "/path/to/uv",
      "args": [
        "--directory",
        "<PATH_TO_CLONED_REPO>",
        "run",
        "server.py",
        "--local"
      ]
    }
  }
}

Optional: grant file-system access

If you want your agent to have file-system access, you can add an MCP filesystem server and point it at the directory you want to allow.

{
  "mcpServers": {
    "file-system": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "~/"]
    }
  }
}

Available tools

deep_search

Plans and breaks down high-level goals into components, identifies suitable MCP servers for each part, and guides the user through decisions when parts are missing.

test_server_template_code

Validates MCP server templates by running test workflows to verify server responses and compatibility during setup.