Streamable HTTP

Monitoring that agents can operate

Streamable HTTP MCP turns the whole monitoring API into tools — create monitors, read status, manage alerts — from any client that can reach a remote server.

{
  "command": "npx",
  "args": [
    "-y",
    "@sdxc/mcp@latest",
    "https://uptime.sergiodxa.com/mcp",
    "--header",
    "Authorization:Bearer ${UPTIME_MCP_TOKEN}"
  ]
}
Agent-first monitoring

The REST API, as tools

Skip writing one-off scripts. Give an agent a scoped API key and let it operate Uptime the same way the REST API does.

  • Every public API capability is a tool — monitors, DNS, TCP, flows, cron jobs, alerts, maintenance, status pages, reports, and team admin.
  • Keys need the mcp scope to connect, then the same resource scopes the REST API already uses.
  • Connect over Streamable HTTP with protocol 2026-07-28 through mcp-bridge on @sdxc/mcp.
Your setup

Connect your client

Create an API key with the mcp scope, add the resource scopes for the tools you want, then paste this mcp-bridge config into Cursor, Claude, or any stdio MCP client.

  1. Create a key

    Open your team’s API keys and create a key.

  2. Grant the scopes

    Grant mcp, plus every other scope the tools you want already require (for example monitors:read).

  3. Add the config

    Paste the mcp-bridge config into your client’s mcp.json.

  4. Ask your agent

    Ask the agent to list monitors or create one — it will call the matching tool.

Put an agent on watch

Sign in, mint a key with mcp plus the resource scopes you need, and paste the config into your client.