clab-mcp-server

Containerlab MCP Server

This is a quick and dirty MCP (Model Context Protocol) trial for interacting with containerlab using AI. The example is tailored for claude desktop.

1. Running containerlab API server
2. Latest build of containerlab from the main branch
   • Build it yourself: make build
   • Copy the binary: cp bin/containerlab $(which containerlab)

Building the MCP Server

For Windows

export GOOS=windows
export GOARCH=amd64
go build -o clab-mcp-server.exe main.go

For Mac

# For Intel Macs
export GOOS=darwin
export GOARCH=amd64
go build -o clab-mcp-server main.go

# For Apple Silicon (M1/M2/M3)
export GOOS=darwin
export GOARCH=arm64
go build -o clab-mcp-server main.go

For Linux

export GOOS=linux
export GOARCH=amd64
go build -o clab-mcp-server main.go

Setup for Claude Desktop

1. Place the MCP server executable in an accessible location
   • For Windows: clab-mcp-server.exe
   • For Mac/Linux: clab-mcp-server
2. Create a configuration file named claude_desktop_config.json with the following content:

# For Windows:
{
  "mcpServers": {
    "clab-api": {
      "command": "C:\\clab-mcp-server.exe",
      "args": [],
      "env": {
        "API_SERVER_URL": "http://localhost:8080"
      }
    }
  }
}

# For Mac/Linux:
{
  "mcpServers": {
    "clab-api": {
      "command": "/path/to/clab-mcp-server",
      "args": [],
      "env": {
        "API_SERVER_URL": "http://localhost:8080"
      }
    }
  }
}

3. Ensure the containerlab API server is running at http://localhost:8080 or update the URL as needed

Usage

Once configured, Claude will be able to:

• List available labs
• Deploy new network topologies
• Inspect lab details
• Execute commands on lab nodes
• Destroy and clean up labs

You can ask Claude to perform these operations using natural language, and it will use the appropriate MCP tools to interact with your containerlab environment.