Let your AI agent run your monitoring: Uptriva now speaks MCP

Setting up monitoring is repetitive: create a status page, add monitors, wire up alert contacts, pick intervals and thresholds. It is also exactly the kind of work an AI agent is good at — structured, API-driven, verifiable.

Today we're shipping an MCP (Model Context Protocol) endpoint built directly into the Uptriva API. Point any MCP-compatible client at it with your API token, and the agent can operate your monitoring workspace through natural language — "watch api.example.com every 60 seconds and alert me on SMS" becomes a few API calls you never had to write.

What the agent can do

The endpoint exposes a focused toolset on top of the same REST API the dashboard uses:

• Monitors — list, inspect, create, update and delete monitors (HTTP, TCP, DNS, ping, SSL, databases, game servers and more)
• Status pages — list, create and inspect public/private status pages
• Incidents — list active incidents, declare new ones, post status updates
• Notification contacts — list and create email/SMS/voice/Slack/Telegram contacts
• Maintenance windows — list and schedule maintenance
• Overview — workspace state at a glance

Every call runs through the normal middleware stack: your token's scopes, plan limits and rate limits apply exactly as if you called the REST API yourself. Nothing is proxied around your permissions.

Setup in one minute

1. Create an API token in the dashboard under Account → API tokens (starts with sk_).
2. Add the remote server to your MCP client:

{
  "mcpServers": {
    "uptriva": {
      "url": "https://api.uptriva.com/api/v1/uptime/mcp",
      "headers": { "Authorization": "Bearer sk_your_token" }
    }
  }
}

That's it. No local install, no extra credentials — the endpoint speaks Streamable HTTP, so it works with Claude Desktop, Cursor and other MCP clients that support remote servers.

A real prompt, start to finish

"Create a status page called 'Acme Prod', monitor https://api.acme.com over HTTPS every 60 seconds, expect status 200, and add an email contact that gets notified on recovery."

The agent calls create_status_page, then create_monitor with type: https, interval: 60, expected_status: 200, then create_notify_contact — and reports the public status page URL back to you. Everything it created shows up in your dashboard like anything you made by hand.

Notes on safety

• Deletion tools are explicit (delete_monitor) — a good agent asks before calling them, and you can revoke the token anytime under Account → API tokens.
• Create a dedicated token for your agent rather than reusing one from a script — that keeps audit trails and revocation clean.
• Read tools (list_*, get_*, get_overview) are side-effect free; let the agent inspect before it writes.

The endpoint and tool list are also discoverable via llms.txt, so agents that crawl the docs find it on their own.

Happy automating.