create_check
Create a check. kind is http, api, dns, tcp, heartbeat, or mcp. Call usage first: it gives min_interval_seconds, max_regions, and max_endpoints (how many active checks the plan allows). Up rules: kind http with expected_status_codes [200] (the default) counts any response below 500 as up, so a 404 or a bot-protection 403 is up; kind api, or any other list of codes, must match exactly. heartbeat takes no target or regions; the result has ping_url, shown only this once, for the job to call with curl -fsS when it finishes (append /fail to report a failure). server (Linux machines with the QuickStatus.dev server agent) takes no target or regions; the result has install_command, shown only this once: run it as root on each Linux server to enroll it. It holds the enrollment token, so do not paste it into files or chat logs. A new server check has rules for CPU and memory above 90% for 10 minutes and any disk above 90%; rules are edited in the console. mcp runs initialize and tools/list against an MCP server URL. Each parameter's description gives its allowed values and default. Request headers, request body, and HTTP auth stay in the console.
Tool call template
Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {
"name": "<name>"
},
"name": "create_check"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| kind | No | http, api, dns, tcp, heartbeat, mcp, or server. | http |
| name | Yes | Display name. Blank comes from the target; a blank heartbeat is named Heartbeat, so pass a name that says which job it watches. | |
| port | No | TCP port (1 to 65535). Required for kind tcp; other kinds reject it. | |
| target | No | http/api/mcp: a full URL (https://...). dns: a domain name. tcp: a hostname or IP only, no scheme or port (pass port separately). heartbeat and server: leave empty. | |
| regions | No | Region codes from list_regions, for example us-west-1. Omit for the default region. At most max_regions from usage. heartbeat and server take none. | |
| site_id | No | Status page id for linking. Pass with site_slug and component_id. | |
| site_slug | No | Status page slug for linking. Pass with site_id and component_id. | |
| dns_server | No | Resolver host or IP, optionally with :port. Empty uses the default resolver. dns only. | |
| verify_ssl | No | Fail the check when the TLS certificate is invalid or expired. Default true. | |
| http_method | No | GET (default), POST, PUT, PATCH, DELETE, or HEAD. http and api only. | |
| min_members | No | server only: the check goes down when fewer machines than this are reporting. Default 1. At most max_server_members from usage (Free 1, Pro 25, Scale 100). | |
| slow_metric | No | total (default: the whole response) or ttfb (time to first byte). | |
| tcp_payload | No | Text to send after connecting. tcp only. | |
| component_id | No | Status page component to link. Pass with site_id and site_slug. | |
| dns_protocol | No | udp (default) or tcp. dns only. | |
| alert_on_lost | No | server only: also warn for each machine that stops reporting without leaving cleanly. Default false. | |
| body_contains | No | Text that must appear in the response body (first 64 KB). http and api only. | |
| incident_mode | No | For a linked check: off, auto (publish incidents), or review (store a draft a person approves before it is published). | |
| tcp_ip_family | No | ipv4 (default) or ipv6. tcp only. | |
| cert_warn_days | No | 7, 14, or 30: warn that many days before an https certificate expires, then once more at each smaller step (14 warns at 14, 7, and 1 day). 0 turns it off. create_check defaults to 14 for https targets; omitted leaves it unchanged on update_check, except that a change to an http:// target clears it and a change from http:// to https:// sets 14. Ignored for http:// targets and for dns and tcp on create; dns and tcp reject it on update. | |
| display_metrics | No | For a linked check: show response time on the public status page. | |
| dns_record_type | No | A (default), AAAA, CNAME, TXT, MX, NS, or CAA. dns only. | |
| timeout_seconds | No | Seconds to wait for a response, 1 to 60. Default 10. | |
| follow_redirects | No | Follow redirects. Default true. http and api only. | |
| interval_seconds | No | Seconds between runs (heartbeat: how often the job pings; server: how often each machine reports, at least 60). At least the plan minimum (min_interval_seconds from usage, for example 300 on Free) and at most 86400. Default: the plan minimum. | |
| slow_threshold_ms | No | http, api, and mcp: a successful run slower than this many ms marks the check slow. The check stays current_status up with degraded_since set, sends a slow alert, and does not count as downtime. 0 turns it off. Off unless set on create_check; omitted leaves it unchanged on update_check. | |
| mcp_expected_tools | No | mcp only: tool names that must appear in tools/list. Tools are listed, never called. | |
| alert_email_enabled | No | Email the workspace owner on down and recovery. Default true. Heartbeats alert after 1 missed or failed ping; other kinds after 3 failed runs in a row, and recover after 1 success. | |
| dns_expected_values | No | Answers that must be present, for example an IP for A or '10 mail.example.com' for MX. dns only. | |
| group_grace_seconds | No | server only: seconds the check may stay below min_members before it goes down, so a reboot or rolling deploy does not alert. Default 300; 0 means at once. | |
| tcp_expected_outcome | No | connect (default: up when the port accepts) or fail (up when it refuses). tcp only. | |
| alert_destination_ids | No | Alert destination ids from list_alert_destinations (Slack, webhook, and so on). | |
| expected_status_codes | No | Status codes that count as up. Default [200]. For kind http, exactly [200] means any response below 500 is up; any other list, such as [200, 201], must match exactly. Kind api always matches exactly. | |
| incident_auto_resolve | No | For a linked check: resolve the incident when the check recovers. | |
| alert_incident_enabled | No | For a linked check: open one status incident per outage (or flapping period). | |
| heartbeat_grace_seconds | No | heartbeat only: extra seconds after the expected ping before it counts as missed. 0 means no grace; up to 86400. Default: the same as interval_seconds. |