AI agent integration

Drumbeats ships an official MCP server so you can run your monitoring straight from your AI client. Point Claude, Cursor, VS Code, or any MCP client at it and create monitors, triage incidents, and run ad-hoc HTTP / SSL / DNS checks — all in plain language. There is nothing to wire into your codebase: the server talks to the Drumbeats API on your behalf.

Use the MCP server when you want to manage monitoring conversationally — "create a cron monitor for my nightly backup", "what's my uptime this month?", "acknowledge the most recent incident" — instead of clicking through the dashboard. The manual language integration guides remain the reference for wiring the start / success / failure pings into your jobs; the MCP server sets up and manages the monitors, your code still sends the pings.

What the agent can do

Through the MCP server an AI client can:

1. Manage projects & monitors. List the projects a key can reach, and — with a key that carries the manage_projects scope — create a new project or update an existing one's name or description. Within a project, create, list, fetch, update, pause, and resume monitors — cron, heartbeat, or HTTP uptime.
2. Observe & triage. Pull a monitor's recent pings and response times, read the project-wide uptime / SLA rollup, list incidents, and acknowledge or resolve them.
3. Run diagnostics with no key. Ad-hoc check_http, check_ssl, and check_dns work without an API key — handy for a quick "is this URL up / is this certificate about to expire?" check.

The server never deletes anything in this release, and key-scoped permissions still apply: an AI client can only do what the API key you give it is allowed to do.

Connecting

1. Mint an MCP key. On the Connect AI page, mint an account-scoped key (dk_…) that works across every project you own. To let the agent create or update projects, give the key the manage_projects scope — monitor and diagnostics tools don't need it.
2. Install the server with one command — npx -y @drumbeats/mcp — and add it to your client's MCP config. The Connect AI page has ready-to-paste config for Claude Desktop, Claude Code, Cursor, and VS Code.
3. Just ask. Once the server is connected, talk to your monitoring in natural language.

The diagnostics tools (check_http / check_ssl / check_dns) run even if you skip the key — omit the DRUMBEATS_API_KEY env var to use the server key-free.

Example prompts

• "List all my monitors and their current status."
• "Create a cron monitor for my nightly backup that runs every day at 02:00 UTC."
• "What's my uptime this month?"
• "Show me open incidents and acknowledge the most recent one."
• "Is https://example.com up right now?" (no key needed)
• "Is the SSL certificate for example.com about to expire?" (no key needed)

Limits

• No deletes. This release of the MCP server exposes no destructive tools — monitors and incidents can be created, updated, paused, resumed, acknowledged, and resolved, but not deleted.
• Scoped to your key. The server can only reach the projects your API key resolves to, and can only perform the operations your key's permission scopes allow.
• It manages monitors, not your code. The MCP server does not edit your repository — it sets up and operates monitors via the API. Your jobs still send the start / success / failure pings; see the language guides for that wiring.

Related guides

• Monitor types — cron, heartbeat, event-driven, and uptime monitors the agent can create.
• Ping API — endpoint reference for the pings your jobs send.
• Production hardening — recommended patterns once your monitors are live.
• Alternatives — how Drumbeats compares to Cronitor and Healthchecks.io.