MCP server

The MCP server authenticates the same way as the HTTP API: an rc_live_ API key sent as a Bearer token. Scopes apply identically — a key with the run scope can execute checks, and a key with the manage scope can create and modify saved checks, scheduled checks, alert channels, and status pages. See API reference for how keys are created and scoped.

The MCP server exposes one tool per API capability, grouped by area:

• Checks & runs — run_check, run_saved_check, list_checks, create_check, get_check, update_check, delete_check.
• Shares — create_share, list_shares, delete_share, get_share_history.
• Scheduled checks & incidents — create_scheduled_check, list_scheduled_checks, get_scheduled_check, update_scheduled_check, delete_scheduled_check, get_scheduled_check_history, list_incidents, get_incident.
• Alert channels — create_alert_channel, list_alert_channels, get_alert_channel, update_alert_channel, delete_alert_channel, test_alert_channel.
• Status pages — create_status_page, list_status_pages, get_status_page, update_status_page, rotate_status_page_token, delete_status_page.
• Other — list_regions, whoami (who the current token is authenticated as), and preview_assertion (test an assertion against a sample response before saving it).

The MCP server is a first-class surface, not an afterthought — it matches the API and web UI for every agent-appropriate action, so an agent can do anything a person can do in the portal for running and managing checks.