Maskbreak System Status

Version 1.3 · Last reviewed 8 September 2026

Automatic checks of the API, pages and database, plus recent server-side outcomes for visitor intelligence, email and sign-in services.

Checking systems...
Fetching live status
Monitored checks
Sample API response
Dashboard page response
Authentication checks
Signup checks
Database read/write
Device intelligence
Network intelligence
Email sending
Google sign-in
GitHub sign-in
Email sign-in

Every minute, checks read and write the database and inspect page and response bodies. The API probe uses /v1/evaluate/sample?scenario=clean; it does not perform a customer's authenticated visitor check. Authentication and signup health checks cover dependencies, not full user journeys or a signed-in dashboard. Cloudflare challenge responses are Unknown, not evidence of an outage. Components monitored by probes, including the database, require two successful probes to confirm recovery.

Provider states reflect real server-side outcomes. Device intelligence uses fingerprint events; network intelligence uses Maskbreak's network intelligence results. Email sending and email/Google/GitHub sign-in reflect observed server-side outcomes. Without recent observations, a service with no unresolved failure is Unknown; customer mistakes such as bad input do not count as service failures. Failures observed in real operations require successful real operations to clear; a passing page probe cannot clear them. Email sending does not confirm inbox delivery, and these checks do not establish that the browser SDK works globally.

GitHub public checks
Independent password check

The periodic dependency checks run inside the application. Separately, GitHub Actions is scheduled every 15 minutes to inspect public page content, the test API contract and reported dependency failures. A password sign-in and account-read check runs only when a dedicated test account is configured; otherwise it stays Unknown. These checks do not create accounts, send verification emails, confirm inbox delivery or complete Google/GitHub sign-in. GitHub schedules can be delayed. Its results become stale after 45 minutes; unresolved failures remain visible until confirmed recovery.

Scheduled public checks
Monitor heartbeat

A separate Cloudflare schedule checks public pages, process health and a simulated API request every five minutes. A missing or stale result after 15 minutes is shown as a monitoring gap, not confirmation of a website outage. Operations receives one notification attempt per incident and one recovery notice after two distinct fresh results, without repeat reminders. Notification attempts are recorded across restarts; unavailable storage or email delivery can prevent a notice. These checks use no account password and do not prove full sign-in or vendor availability. Open the independent status snapshot if this page is unavailable. It runs outside Railway but shares Cloudflare with the main site. The existing external uptime alarm remains separate.

If the application stops, this page and its incident history may be unreachable. The independent snapshot can still show its latest result, marked Unknown after 15 minutes without an update. Contact support@maskbreak.com if you cannot access the service. These checks are not global availability measurements; gaps and untested failures can be missed.

Recorded probe samples — Last 90 Days
90 days agoToday
Recorded samples where all three HTTP probes passed. This historical percentage covers the sample API, dashboard and login HTTP checks only; it excludes the other service indicators above. Before 8 September 2026, dashboard and login checks covered HTTP page responses and could accept a challenge page. From 8 September, HTTP probes also validate response bodies, and combined samples are omitted whenever any page probe is Unknown. Earlier samples do not establish that body validation passed. Database and authentication dependency checks affect live service states, not these historical percentages. Sample success is not complete uptime or a contractual SLA measurement. Operating aims: service level objectives.

Loading recorded history. Grey days have no usable samples.

Decision Latency — Measured
Loading measured latency…

Server-side processing only — the time Maskbreak spends producing a verdict. It excludes your network round-trip. Live percentiles use the latest recorded decisions in this process (up to 1,000 per surface); the historical figures combine recorded evaluation and lookup decisions. They are not global edge measurements or proof that a monthly aim was met. Any millisecond figure beside Sample API response is the average duration of successful probes, including that probe's network path. Before 8 September 2026, probe timing ended at response headers; newer checks inspect response bodies. Neither measures a real visitor evaluation.

Incidents & Maintenance

Automatic incidents are retained for 90 days alongside curated historical incidents and maintenance entries. An empty log, or no entry for a period, is not proof that no outages occurred. During a database outage, current status stays in memory and pending incidents are buffered, then saved on recovery. A restart before the database recovers can lose that buffered history. Raw hosting logs are separate from this incident log. Notice periods, if agreed, are set out in your signed agreement.

Loading incident history.

Loading incident log…
Process running time:
Last probe:
Infrastructure: Railway + Cloudflare edge
← Back to Maskbreak Log in →