Connect Claude, Cursor or any MCP client to Uptriva's built-in MCP endpoint and let an agent manage monitors, incidents and alerting with your API token.

Uptriva exposes a remote MCP (Model Context Protocol) server built into the uptime API. MCP-compatible clients — Claude Desktop, Cursor, and others — can call it to manage your monitoring workspace through structured tools instead of raw HTTP requests.

Endpoint

POST https://api.uptriva.com/api/v1/uptime/mcp

Transport: Streamable HTTP (JSON-RPC 2.0, one request per POST)
Auth: Authorization: Bearer sk_your_api_token — the same personal API tokens used for the REST API
Token scopes, plan limits and rate limits apply to tool calls exactly as they do to REST calls

Client configuration

Claude Desktop / Cursor mcp config:

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

Create a token in the dashboard under Account → API tokens. Use a dedicated token per agent so you can revoke it independently.

Available tools

Monitors

Tool	Purpose
`list_monitors`	List monitors, optionally filtered by `status_page_id`
`get_monitor`	Fetch one monitor with status and config
`create_monitor`	Create a monitor (`name`, `type`, `target` required; `interval`, `expected_status`, `http_method`, `failure_threshold`, `is_internal` optional)
`update_monitor`	Update fields on an existing monitor
`delete_monitor`	Permanently delete a monitor — destructive

Monitor types: http, https, ping, icmp, tcp, udp, dns, smtp, pop3, imap, postgres, mysql, redis, telnet, ftp, sftp, ssh, ssl, domain, heartbeat.

Status pages

Tool	Purpose
`list_status_pages`	List your status pages
`get_status_page`	Fetch one status page
`create_status_page`	Create a page (`name` required; `slug`, `description`, `is_public` optional)

Incidents

Tool	Purpose
`list_incidents`	List incidents, optionally filtered by `status_page_id`
`get_incident`	Fetch one incident
`create_incident`	Declare an incident (`status_page_id`, `title` required; `severity`, `status`, `monitor_ids`, `is_public` optional)
`add_incident_update`	Post a status update to an incident

Severity: minor, major, critical. Status: investigating, identified, monitoring, resolved.

Notification contacts

Tool	Purpose
`list_notify_contacts`	List contacts for a status page (`status_page_id` required)
`create_notify_contact`	Create a contact — `email`, `slack_webhook_url` or Telegram required; `sms`/`phone` in E.164; `notify_on_recovery` optional

Maintenance

Tool	Purpose
`list_maintenance_windows`	List scheduled maintenance
`create_maintenance_window`	Schedule maintenance (`status_page_id`, `title`, `starts_at`, `ends_at` — RFC3339 timestamps)

Workspace

Tool	Purpose
`get_overview`	Monitor counts by status and recent incidents

Recommended workflow for agents

list_status_pages → find or create_status_page the target page
create_monitor for each endpoint to watch
list_notify_contacts / create_notify_contact for alert delivery
get_overview to verify the final state

Read tools (list_*, get_*, get_overview) have no side effects. Write tools map 1:1 to REST endpoints — everything an agent creates behaves exactly like dashboard-created resources and shows up in the UI.

Notes

delete_monitor is irreversible — agents should confirm with the user before calling it.
Tool errors are returned as MCP isError results containing the API's HTTP status and error body, so agents can recover gracefully (e.g. validation failures come back as readable messages).
The endpoint is advertised in /llms.txt for agent discovery.

Use Uptriva with MCP (AI agents)