io.github.ronniechong/healthchecks-io-mcp

@digitalronin/healthchecks-io-mcp

Unofficial, unaffiliated with Healthchecks.io. This is a third-party MCP server, not an official Healthchecks.io product.

What this is

Healthchecks.io is a "dead man's switch" style monitoring service: your scheduled jobs (cron jobs, backups, batch scripts, anything that's supposed to run on a schedule) ping it when they run, and Healthchecks.io alerts you if a ping doesn't show up on time — meaning the job silently failed or never ran.

This package is an MCP server — a small local program that lets an AI assistant like Claude talk to Healthchecks.io's Management API on your behalf. Once it's set up, you can ask your AI assistant things like "list my Healthchecks.io checks," "show me the ping history for my backup job," or "pause monitoring for my staging environment" in plain English, and it will call the right Healthchecks.io API endpoint for you.

Setup

1. Get a Healthchecks.io API key

1. Log in to healthchecks.io (or your self-hosted instance, see below).
2. Go to your project's Settings page, then the API Access tab.
3. You'll see two keys: a read-only key and a read-write key.
   • The read-only key can look up your checks, but can't create, change, pause, or delete anything.
   • The read-write key can do everything the read-only key can, plus create, update, pause, resume, and permanently delete checks.

2. Add this server to your MCP client's config

You don't need to install anything manually — npx will download and run it automatically the first time it's used. Add this block to your MCP client's configuration (for Claude Code, this is a .mcp.json file; other clients have their own config file or UI for this):

{
  "mcpServers": {
    "healthchecks-io": {
      "command": "npx",
      "args": ["@digitalronin/healthchecks-io-mcp"],
      "env": {
        "HEALTHCHECKS_API_KEY": "your-key-here"
      }
    }
  }
}

Replace "your-key-here" with the API key from step 1. Restart your MCP client (or reload its MCP connections) after saving this — most clients only pick up new/changed servers on restart.

3. Try it out

Once connected, just ask your AI assistant something like:

• "List all my Healthchecks.io checks"
• "Show me the ping history for my nightly-backup check"
• "Is my staging-cron check currently failing?"

If it responds with real data from your account, you're set up correctly.

Optional: multiple projects

Healthchecks.io API keys are scoped per project, so a single server entry only ever talks to one project. To work with more than one, register the server multiple times under different names, each with its own key:

{
  "mcpServers": {
    "healthchecks-io-personal": {
      "command": "npx",
      "args": ["@digitalronin/healthchecks-io-mcp"],
      "env": {
        "HEALTHCHECKS_API_KEY": "personal-project-key-here"
      }
    },
    "healthchecks-io-work": {
      "command": "npx",
      "args": ["@digitalronin/healthchecks-io-mcp"],
      "env": {
        "HEALTHCHECKS_API_KEY": "work-project-key-here"
      }
    }
  }
}

Each entry runs its own process holding a single key, and MCP clients namespace tools by server name, so you can tell which project a tool call is hitting.

Optional: self-hosted Healthchecks.io

Healthchecks.io is open source, and some people run their own instance instead of using the hosted SaaS service at healthchecks.io. If that's you, add a second environment variable pointing at your instance's API root:

"env": {
  "HEALTHCHECKS_API_KEY": "your-key-here",
  "HEALTHCHECKS_BASE_URL": "https://monitoring.example.com/api/v3"
}

The URL must be the full API root, including the /api/v3 path segment, with no trailing slash — e.g. https://monitoring.example.com/api/v3, not https://monitoring.example.com or https://monitoring.example.com/api/v3/. Getting this wrong will make every tool call fail with a "not found" error, since the server appends paths like /checks/ directly onto whatever you set here. If you leave HEALTHCHECKS_BASE_URL unset, it defaults to the real https://healthchecks.io/api/v3.

If your self-hosted instance isn't behind HTTPS, the server will print a warning (but still run) — your API key is sent as a request header on every call, so an http:// URL means that key travels in plaintext over the network to your instance.

What it can do

This server exposes 11 "tools" that your AI assistant can call. They fall into two groups:

Read tools — safe, look-up-only, never change anything:

Mutating tools — these change things in your account, and all of them require a read-write API key:

Technical reference

For each tool, this table gives its underlying Healthchecks.io API endpoint: