Every release,
in the open.

• NewAlert rules (API) now honor a schedule window: set a start time, end time, and timezone on a rule and it will only alert inside that window, including overnight windows like 22:00 to 06:00. Rules without a schedule behave as before.
• ImprovedSchedule times and timezones on alert rules are now validated when you create or update a rule, so a typo cannot silently disable alerting.

• FixPage authentication settings for visual monitoring now save correctly from the monitor Settings panel. Previously, editing settings there could break a configured page login. Saved credentials are kept when you leave a credential field blank while editing.
• FixThe alert delay and report settings chosen while creating a monitor are now applied to the new monitor. Previously they only took effect when set again after creation.
• SecurityEnhanced security hardening.

• FixThe AI usage section on the Billing page now reports your actual AI activity (chat assistant, incident summaries, digests, alert enhancements) with usage and cost per feature. Previously it always showed zero.

• FixReplying STOP to a WhatsApp alert now reliably stops alerts to that number, with a confirmation message. Reply START to turn them back on. Other numbers on the same alert channel are not affected.

• FixThe record-type checkboxes on DNS monitoring now control exactly which record types are watched. Previously all record types were monitored regardless of your selection, so a deselected type could still trigger a change alert.
• ImprovedNarrowing your record-type selection no longer triggers a one-time false change alert for the types you removed.

• FixThe LCP, First Contentful Paint, and Layout Shift cards on the Performance tab now show real values from the daily page speed test. Previously they could stay empty even when the data was available.
• ImprovedThe DNS Lookup card now shows an instant result when your domain resolves immediately, instead of appearing empty.
• ImprovedScheduled email reports now have their own link in the sidebar and on the settings page, so digest reports are easy to find and set up.
• ImprovedThe page speed section now shows which device profile the test ran on.

Editing many monitors at once now applies exactly what you chose, tells you exactly what happened, and looks right in dark mode.

• FixBulk edit now reports precisely how many monitors were updated, and tells you why any were skipped (for example, a setting not available on your plan). Previously it could report success even when a change was not applied.
• FixBulk-editing one visual monitoring setting no longer resets the others. Changing just the threshold, screenshot size, or full-page option now leaves your remaining visual settings untouched.
• NewDomain expiry alerts can now be turned on or off for many monitors at once, alongside visual monitoring and SSL checks.
• ImprovedScreenshot width and height can now be changed independently in bulk edit. Previously picking only one of the two was quietly ignored.
• FixThe domain expiry toggle in monitor settings now saves reliably.
• ImprovedTurning visual monitoring on in bulk applies to website monitors and clearly skips monitor types where screenshots do not apply.
• FixDropdown menus in the bulk edit dialog now display correctly in dark mode.

A thorough pass over API monitoring: response checks now run consistently from every region, API Key authentication works, and the settings you configure actually take effect.

• FixResponse assertions, expected content, and keyword checks now run from every monitoring region, not just Europe. Previously a check run from a non-European region validated only the status code, so a bad response could pass and a real failure could be missed.
• FixAPI monitors that send a request body (including GraphQL) are now checked with their full request from every region, so endpoints that require a body are tested correctly everywhere.
• FixAPI Key authentication now sends your key on every check, and saving an API Key auth configuration no longer fails.
• FixResponse assertions now apply to GraphQL monitors, and an assertion like "no errors present" now correctly flags a 200 response that carries an error payload.
• FixA monitor that starts returning an empty response, or a status code outside the ones you marked as expected, is now correctly flagged instead of quietly passing.
• ImprovedThe Response SLO and maximum response size you set on an API monitor now take effect: a slow response past your SLO is flagged, and larger response bodies are captured up to the size you choose.
• FixNotification channels chosen in an API monitor's Alert settings now save correctly and show your real selection when you reopen the section.
• ImprovedAssertion-failure alerts to Slack and Telegram are now reliably delivered even when the response value is large.
• ImprovedThe API monitors dashboard now shows GraphQL monitors, loads faster on large accounts, and displays the correct request method, authentication indicator, and response time on each card.
• ImprovedAPI monitor timeouts can now be set up to 120 seconds, matching what the form offers.
• SecurityEnhanced security hardening.

• FixThe alerts feed (dashboard overview and the mobile app Activity tab) could show as empty even when incidents had occurred. It now includes website down and recovery events plus server alerts, so your full alerting history is visible.

• FixSaved API bearer tokens are now retained when you edit other monitor settings after the token field is masked.

Outage detection is no longer tied to how often your monitor runs. Once a failure is confirmed, VisualSentinel rechecks on its own fast cadence until the alert decision is made.

• NewAfter the first confirmed failure, your site is automatically rechecked about once a minute (from alternating regions where available) until it recovers or the alert threshold is reached. A monitor checked every 10 minutes now alerts in about 2 to 3 minutes instead of waiting out multiple full check intervals.
• ImprovedThe consecutive-failures alert setting now means exactly what it says: the number of confirmed failed checks before we alert you. Previously the alert could fire earlier than the configured count.
• ImprovedMonitors checked from multiple regions alert immediately when every selected region independently confirms the site is down.
• ImprovedAlert delay timing is now measured from when the failure streak actually started, making the delay setting accurate at every check interval.

• FixFixed the announcement bar overlapping the top navigation on phones.

• FixFixed alerts to Discord not being delivered for accounts using a custom uploaded logo. Alerts for affected accounts now arrive normally.

• ImprovedWhen several sites on the same linked server slow down together, you now get a couple of alerts and then quiet instead of a separate message for every site, every few minutes.
• ImprovedA site that keeps flipping between slow and normal now alerts once per episode instead of on every flip. Recovery messages only arrive for slowdowns you were actually alerted about.

• FixFixed a rare case where a brief system hiccup could show an incorrect "limit reached" message or temporarily hide features included in your plan.
• ImprovedVisual checks now run even if an internal plan lookup briefly fails, so short-lived hiccups on our side can never skip a scheduled visual check.

• FixFixed a case where a failed visual check could stop later visual checks for the same site from running, so visual changes are now captured more reliably.

• FixFixed a case where scheduled visual checks could stop running for some monitors, causing visual changes to be missed until another trigger happened.
• FixImproved visual-change incident creation so a detected visual difference is less likely to be left without the matching incident record and alert.

• FixFixed the "Default" badge on notification channels being unreadable in light mode.

• FixFixed a case where an alert to your Discord channel could silently fail to send. Alerts now always post, even when some details are missing.

• FixMonitor cards now reflow to fewer columns on narrower screens instead of cramming into thin, truncated cards, and the top navigation no longer wraps or trims at medium widths.