UptimeDuck for AI agents

A short reference for AI assistants and agents working with UptimeDuck on someone's behalf. Setting this up yourself? Start with the MCP server docs.

What UptimeDuck does

UptimeDuck monitors websites, APIs, TCP ports, ping targets, Minecraft servers and DNS blacklists from multiple regions. A small agent reports CPU, RAM and disk for Linux servers and Kubernetes clusters, and cron monitors watch scheduled jobs through heartbeats. When something goes down, UptimeDuck alerts by email, mobile push, SMS, Slack, Discord, Telegram or Matrix, and projects can have a public status page.

Every account starts on the Free plan. Paid plans and their limits are on the pricing page and in Plans & limits.

Connect over MCP

• Prefer sign-in. The user approves the connection in their browser and picks read only or read & write, so they never have to paste a token into a chat.
• Call whoami first. It returns the account, workspace, plan and scope you are acting with.
• API access is available on every plan, Free included. Each plan limits how many API tokens the user can have at the same time; see Plans & limits.

Good to know

• create_heartbeat_monitor creates the monitor paused, like the panel does. Give the user the push URL, and call resume_monitor once their job is sending heartbeats. Otherwise the monitor would go down before the first run.
• delete_monitor needs confirm: true. Ask the user before deleting, since the monitor's history is deleted too.
• Write tools are refused with FORBIDDEN on read only access, and with PLAN_LIMIT_REACHED when the plan's limit on monitors, intervals, regions or alert channels is reached. Nothing is changed in either case.
• Each token can make 120 requests per minute by default. Above that, requests get HTTP 429.
• Every change is recorded with the token or connected app that made it, and the user can disconnect you at any time.

Without MCP

• REST API: the same API tokens work with https://api.uptimeduck.com/api/v1, with the same scopes.
• Heartbeats: a job reports in with curl "https://api.uptimeduck.com/push/MONITOR_ID?status=up&msg=OK". No API key is needed. See Cron & heartbeat API.
• Linux servers: curl -sL https://uptimeduck.com/download/agent/linux/install.sh | sudo bash -s -- --token=YOUR_SERVER_TOKEN, with the token from the server monitor in the panel. See Server agent.
• Kubernetes: Helm chart uptimeduck/uptimeduck-agent from https://uptimeduck.com/charts. See Kubernetes (Helm).

Key URLs

Questions? Email [email protected].