Status

System status

A scripted client connects to the public MCP endpoint every five minutes, lists the tools, lists an inbox and reads a message. Everything below is counted from what those checks recorded.

All systems operational

The most recent check passed.

Last check Sep 8, 2026, 20:25 UTC

Uptime, last 30 days

99.72%

8,596 checks, one every 5 minutes. 24 of them failed.

Continuous monitoring started Aug 3, 2026.

Last 30 days

  • Aug 10, 2026: 288 checks, all passed
  • Aug 11, 2026: 288 checks, all passed
  • Aug 12, 2026: 288 checks, all passed
  • Aug 13, 2026: 288 checks, all passed
  • Aug 14, 2026: 288 checks, 1 failed
  • Aug 15, 2026: 288 checks, all passed
  • Aug 16, 2026: 288 checks, 1 failed
  • Aug 17, 2026: 288 checks, 1 failed
  • Aug 18, 2026: 287 checks, 3 failed
  • Aug 19, 2026: 288 checks, 1 failed
  • Aug 20, 2026: 288 checks, all passed
  • Aug 21, 2026: 288 checks, all passed
  • Aug 22, 2026: 288 checks, all passed
  • Aug 23, 2026: 288 checks, all passed
  • Aug 24, 2026: 288 checks, 1 failed
  • Aug 25, 2026: 287 checks, 3 failed
  • Aug 26, 2026: 288 checks, 2 failed
  • Aug 27, 2026: 288 checks, 1 failed
  • Aug 28, 2026: 288 checks, all passed
  • Aug 29, 2026: 288 checks, all passed
  • Aug 30, 2026: 288 checks, 1 failed
  • Aug 31, 2026: 288 checks, all passed
  • Sep 1, 2026: 288 checks, all passed
  • Sep 2, 2026: 288 checks, all passed
  • Sep 3, 2026: 288 checks, 1 failed
  • Sep 4, 2026: 288 checks, all passed
  • Sep 5, 2026: 288 checks, all passed
  • Sep 6, 2026: 288 checks, 6 failed
  • Sep 7, 2026: 288 checks, 1 failed
  • Sep 8, 2026: 246 checks, 1 failed
30 days agoToday
All checks passedOne check recoveredSome checks failedNo data

What is checked

One pass through the four steps, in this order. A run stops at its first failure, so each step is scored only against the checks that reached it.

Incidents

A single failed check that passes again five minutes later is counted against uptime above but is not listed here as an incident. That is the same line the pager draws: two consecutive failures, or one failure of a kind that cannot be somebody else's network.

  1. Sep 6, 2026, 08:00 UTC

    The upstream mail provider did not answer in time

    3 consecutive checks failed. Affected: Mail read. Recovered Sep 6, 2026, 08:15 UTC.

  2. Aug 24, 2026, 18:25 UTC

    A malformed MCP response

    1 check failed. Affected: Connection. Recovered Aug 24, 2026, 18:30 UTC. Paged on the first failure.

20 single failed checks in this window, each of which passed again on the next run.

What this page does and does not tell you

  • The check runs from one place on a five minute cron, so an outage shorter than five minutes can pass unrecorded, and the timestamps here are accurate to the check that caught it rather than to the second.
  • It exercises four code paths. Sending, scheduling, drafts and folder operations are not on this page because nothing measures them on a schedule, and a figure covering them would be a guess.
  • It reads one real mailbox at one provider. A failure attributed to the mail provider is that provider having a bad minute, not necessarily anything you would have noticed on yours.
  • Nothing here is derived from customer traffic. No account, workspace or message is counted, sampled or shown.

Every number on this page is counted from the monitor's own recorded runs at read time. Nothing is entered by hand.

How we handle your mail and credentials

Something wrong that this page says is fine? Write to hello@mcpemails.com.