Log in Sign up

AI agents and MCP

Let AI agents check sites, SSL and domain expiry and manage monitors: the MCP server, the agent skill, the free tools API, OpenAPI and llms.txt.

AI agents, chat assistants and scripts can use uptime.rs in four ways. The checks (website, SSL, domain expiry, ping) need no account; managing monitors needs an API key.

MCP server

The Model Context Protocol server is at https://uptime.rs/mcp (streamable HTTP; answers are JSON). Add it as a remote MCP server in Claude, Cursor, VS Code or any MCP client. To use the monitor tools, add the header Authorization: Bearer <your API key>.

{
  "mcpServers": {
    "uptime-rs": {
      "type": "http",
      "url": "https://uptime.rs/mcp",
      "headers": { "Authorization": "Bearer urs_…" }
    }
  }
}
ToolWhat it doesKey
check_websiteRequests a URL now: up or down, status code, response time, redirects, server, Cloudflare, challengeNo
check_sslThe certificate: issuer, names, expiry, days left, trusted by browsersNo
check_domain_expiryRegistration, expiry date, days left and registrar (RDAP or WHOIS)No
pingFour TCP connects to host:port (443 by default) with min/avg/max timeNo
list_monitors, get_monitorYour monitors with status and uptime; one monitor with 24h/7d/30d uptime and each location's latest resultRead
monitor_checks, list_incidentsLatest checks of a monitor; incidents of a monitor or of the accountRead
add_monitors, pause_monitor, resume_monitorAdd targets, pause and resume monitorsWrite

Monitor tools count against the key's limits, like API requests. Check tools allow 30 calls per minute per IP address.

Agent skill

A skill tells an agent when to use uptime.rs and how to read the answers. Download uptime-rs-skill.zip (or read SKILL.md) and add it to your agent: in Claude, under Settings → Capabilities → Skills; in Claude Code, unzip it into ~/.claude/skills/. The skill uses the free tools API below, so it works without MCP.

Free tools API

Plain GET requests with no key, answering JSON in data. Pass the site as host (url and domain also work). Limit: 30 requests per minute per IP address.

curl "https://uptime.rs/api/v1/tools/website?host=example.com"
curl "https://uptime.rs/api/v1/tools/ssl?host=example.com"
curl "https://uptime.rs/api/v1/tools/domain?host=example.com"
curl "https://uptime.rs/api/v1/tools/ping?host=example.com&port=443"
{
  "data": {
    "ok": true,
    "host": "example.com",
    "port": 443,
    "issuer": "Let's Encrypt",
    "expires_at": "2027-01-12T08:14:03+00:00",
    "days_left": 76,
    "trusted": true,
    "error": null,
    "checked_at": "2026-10-28T12:00:00+00:00"
  }
}

Answers are cached briefly (website 1 minute, SSL 5 minutes, domain 6 hours, ping 30 seconds). Addresses in private networks are refused. The same checks are on the web as free tools.

OpenAPI and llms.txt

  • /openapi.json describes the free tools API and the REST API (OpenAPI 3.1), for agents and API clients that read it.
  • /llms.txt summarises the site and links every feature, tool and docs page for language models.