upok.dev — uptime checks with email alerts

GETs a URL every 10 minutes; 2xx = up. Emails you when a check goes down and
when it comes back. Every check has a public status page. No signup.

QUICKSTART

  1. Key:     curl -X POST https://upok.dev/v1/keys
  2. Check:   curl -X POST https://upok.dev/v1/checks -H "Authorization: Bearer KEY" \
                -H "Content-Type: application/json" \
                -d '{"url":"https://api.example.com/health","email":"you@example.com"}'
  3. Status:  curl https://upok.dev/v1/checks/ID -H "Authorization: Bearer KEY"

ENDPOINTS

  (auth) = "Authorization: Bearer upok_...". Errors: {"error": "..."} with the
  matching HTTP status. Timestamps: epoch ms.

  POST   /v1/keys                 -> 201 {"key"}. Never stored; keep it.
  POST   /v1/checks        (auth) body {"url", "email", "name"?}  (name <= 80 chars)
                                  -> 201 check | 400 invalid | 429 over 100 checks
  GET    /v1/checks        (auth) -> {"checks": [check, ...]}
  GET    /v1/checks/:id    (auth) -> check | 404 if not yours
  DELETE /v1/checks/:id    (auth) -> 204
  GET    /c/:id.json              -> check without email fields, plus
                                     "incidents": [{started_at, ended_at|null, duration_ms|null, error}]
                                     "recent": [{ts, ok, latency_ms, status}]  (last 144)
  GET    /c/:id                   status page (HTML)
  GET    /llms.txt                this document

  No auth; reached from mails and the landing form:
  GET    /confirm/:account/:token  asks first; POST to the same URL enables alerts
  GET    /c/:id/delete/:token      asks first; never deletes on GET
  POST   /c/:id/delete/:token      deletes the check
  POST   /new                      form (url, email): key + check in one step

  check = {
    "id", "name", "url", "email", "email_confirmed",
    "status": "pending" | "up" | "down",
    "since": <ms, start of the current up/down run>,
    "last_checked_at", "last_status", "last_latency_ms", "last_error",
    "uptime": {"24h", "7d", "30d"}   (ratio of ok results, null without data)
    "interval_seconds": 600, "status_url", "created_at"
  }

RULES

  - Every 10 minutes, 10 s timeout, redirects followed. 2xx = up; anything
    else is retried once and the retry is what gets recorded.
  - Mail only on change: one when down, one when back up.
  - Alerts need the address confirmed once per key. The first check for an
    address sends a confirmation link; until it is confirmed, checks run but no
    mail is sent (email_confirmed: false).
  - Alert mails carry the status page link and a delete link.
  - History: 30 days.