Mackerel

You run the Mackerel MCP Server to expose a programmable interface for managing hosts, services, metrics, monitors, alerts, downtimes, and notification channels in Mackerel. It lets you control these resources remotely through MCP-compliant clients, enabling automation and centralized monitoring workflows.

How to use

You connect an MCP client to the Mackerel MCP Server to perform host and service management, post and fetch metrics, create and update monitors, close alerts, and manage downtimes and notification channels. Use the provided MCP endpoint through a standard client that speaks MCP RPCs. You will typically authenticate with your Mackerel API key, and operations will reflect in your Mackerel account. Treat the server as your programmable control plane for Mackerel resources, enabling automated workflows and integration with your CI/CD, incident response, and monitoring dashboards.

How to install

{
  "mcpServers": {
    "mackerel_mcp": {
      "command": "uv",
      "args": [
        "run",
        "--directory",
        "/app/mackerel_mcp_server",
        "server.py"
      ],
      "env": {
        "MACKEREL_API_KEY": "YOUR_MACKEREL_API_KEY"
      }
    }
  }
}

Prerequisites: you need Python 3.12 or newer and a Mackerel API key. Install a runtime for the MCP server as shown in the configuration block above, which uses the uv runner to launch the Python server from its directory. Ensure your environment has access to the Mackerel API and that the API key is securely provided as an environment variable.

Step-by-step setup:

• Install Python 3.12 or newer on your host.
• Obtain a Mackerel API key and keep it secure.
• Prepare the MCP server configuration block shown above and place it in your deployment environment so the server can launch with the API key set in MACKEREL_API_KEY.
• Start the MCP server using the specified runtime command: uv run --directory /app/mackerel_mcp_server server.py.

Additional sections

Configuration notes: The server is implemented in Python and expects a Mackerel API key to authenticate against the Mackerel API. The configuration example uses a local runtime command to start the server, so adapt the path /app/mackerel_mcp_server to your deployment when hosting in containers or on your server.

Available tools

Host management

List hosts, fetch host details, update host status, and retire hosts to remove them from active monitoring.

Service management

List and fetch service data, and manage service roles within MCP clients and automations.

Metrics (post/get)

Post host and service metrics and query metric data to drive dashboards and alerts.

Monitors

Create, update, list, and delete monitors to observe defined conditions across your environment.

Alerts

List alerts and close them when resolved or acknowledged.

Downtimes

List, create, update, and delete downtimes to suppress alerts during planned maintenance.

Notification channels

List, create, and delete channels used to deliver alerts to Slack, email, or other endpoints.