UptimeEye MCP server

Quick start

1. Create a management key

In the app open API Keys → New API Key, choose the type “Management” and copy the key. It starts with ue_live_ and is shown once. Log ingest keys (ue_ingest_) do not work here.

2. Add the server to your client

Endpoint: https://api.uptimeeye.com/mcp (streamable HTTP), authenticated with Authorization: Bearer ue_live_…. With Claude Code it is one command:

claude mcp add --transport http uptimeeye https://api.uptimeeye.com/mcp \
  --header "Authorization: Bearer ue_live_XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX"

3. Ask

Try “List my UptimeEye monitors” — then describe what you want monitored. The assistant looks up your notification channels first, because every monitor needs one.

Connect your client

Claude Code

The quick start command adds the server for you alone. Add --scope project to share it with your team via .mcp.json, and keep the key out of that file: in single quotes, ${UPTIMEEYE_API_KEY} goes into the file as it is, and Claude Code reads the key from that environment variable when it starts.

claude mcp add --transport http --scope project uptimeeye https://api.uptimeeye.com/mcp \
  --header 'Authorization: Bearer ${UPTIMEEYE_API_KEY}'

Cursor

Global in ~/.cursor/mcp.json, or per project in .cursor/mcp.json:

{
  "mcpServers": {
    "uptimeeye": {
      "url": "https://api.uptimeeye.com/mcp",
      "headers": { "Authorization": "Bearer ue_live_XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX" }
    }
  }
}

VS Code (GitHub Copilot agent mode)

VS Code prompts for the key once and stores it securely:

{
  "inputs": [
    { "type": "promptString", "id": "uptimeeye-key", "description": "UptimeEye API key (ue_live_…)", "password": true }
  ],
  "servers": {
    "uptimeeye": {
      "type": "http",
      "url": "https://api.uptimeeye.com/mcp",
      "headers": { "Authorization": "Bearer ${input:uptimeeye-key}" }
    }
  }
}

Claude Desktop

Claude Desktop starts local servers only, so it connects through the mcp-remote bridge (needs Node.js). Add this to claude_desktop_config.json and restart the app:

{
  "mcpServers": {
    "uptimeeye": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://api.uptimeeye.com/mcp", "--header", "Authorization:${UPTIMEEYE_AUTH}"],
      "env": { "UPTIMEEYE_AUTH": "Bearer ue_live_XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX" }
    }
  }
}

Other clients: anything that supports remote MCP servers over streamable HTTP with a custom header works with the endpoint and header above.

Available tools

Tools carry MCP annotations: read-only tools are marked as such, delete tools as destructive, so well-behaved clients ask before running them.

Scheduled tasks are pinged by your job: GET https://api.uptimeeye.com/v1/beat/{taskId} on success. Optional: /start when the run begins and /fail to report a failed run and alert immediately. GET, POST and HEAD are all accepted.

Logs tools

Five read-only tools search the logs your services send to UptimeEye Logs. Queries are LogsQL, e.g. service:api level:error. The time range is set with start and end (RFC 3339) and defaults to the last hour; every result states the range actually searched, which your plan's retention can shorten.

Example prompts

• “Monitor https://shop.example.com every minute from eu-central-1 and us-east-1, alert me by email at ops@example.com if it's down or slower than 2 seconds.”
• “Create a heartbeat for our nightly backup at 02:30 Europe/Berlin with 30 minutes grace and give me the curl line for the crontab.”
• “Alert me if our Postgres at db.example.com:5432 stops accepting connections — check every minute.”
• “Which monitors are currently down, and are there open incidents?”
• “The checkout monitor went down at 14:05. What did the checkout service log around then, and is that error new?”
• “Tag every monitor whose URL contains staging with staging and pause them.”
• “Move all monitors from the ‘Old alerts’ channel to ‘On-call’, then delete ‘Old alerts’.”

Set up a project with an AI agent

Coding agents (Claude Code, Cursor, Copilot) can read your repository and the UptimeEye docs at the same time. Paste one of these prompts into the agent in your project; it finds the endpoints, jobs and stack by itself.

Monitor every public endpoint

Add heartbeats to cron jobs

Ship logs

Permissions & safety

• A management key acts on one organization with the same rights as the REST API. Create one key per client so you can revoke them separately.
• A management key can read all of the organization's logs, within your plan's retention, through the logs tools and the REST API. There is no narrower log permission, so treat the key like the logs themselves.
• Delete tools are annotated as destructive and the server instructs the assistant to ask for explicit confirmation first.
• Plan limits apply exactly as in the app — the assistant cannot exceed your monitor count or minimum interval.
• Keys are stored hashed; the plain key is shown once at creation. Requests are processed in the EU (Germany).

Prefer infrastructure as code? The same operations are available through the Terraform provider.

FAQ