Every release,
in the open.

If you have many monitors on the same hosting (even without explicitly grouping them), a shared infrastructure event would previously page you for every site at once. We now spot the pattern automatically and consolidate to one informative alert.

• ImprovedWhen several of your monitors that share the same internet address fail at the same time, we now collapse the notifications into one alert that names the shared address. The per-monitor incidents still appear on your dashboard, but you stop getting flooded.
• ImprovedDetection is automatic and per-account: even if monitors share a public CDN address with someone else, the alert grouping never crosses account boundaries.

When many of your monitors share infrastructure that goes wobbly, we already fold the failure alerts into one summary. We now do the same for slow-response alerts so you stop getting a flood of "responding slowly" notifications during a shared-host event.

• ImprovedSlow-response alerts are folded into the main outage summary when many monitors on the same hosting are slowing down at once. The per-monitor incident still appears on your dashboard so nothing is hidden, but you get one consolidated alert instead of dozens.

Hotfix for an internal background process. No user-visible changes.

• FixResolved a background error that prevented one of our internal correlation checks from running. No customer-facing behavior changes.

Performance and observability follow-ups based on real user data from the previous release.

• ImprovedVisual monitoring page is dramatically faster. Layout no longer shifts while content loads, clicks respond instantly, and large accounts load progressively instead of all at once.
• FixInternal error tracking now captures real stack traces instead of opaque "Script error." messages, so we can fix issues faster.

Foundational changes to how we decide when to wake you up. When our checking infrastructure is itself unstable, we now hold customer alerts until we can confirm the issue is on the monitored site, not on our side. Phase 1 of a multi-step rollout.

• ImprovedInternal trust signals so we can tell "our infrastructure is having a moment" apart from "your site is actually down." First customer-visible benefit lands in a follow-up release.
• ImprovedResponse-time charts no longer get ruined by a single slow check. One outlier used to stretch the Y-axis so far that the typical-case variation became a flat line. Charts now resist outliers while still showing the spike if you hover.

Small follow-up to the previous release. No user-visible changes.

• FixQuieter handling of malformed requests to public endpoints. No customer-visible behavior change.

Small follow-up to the previous release. No user-visible changes.

• FixInternal observability instrumentation now activates correctly in production. No customer-visible behavior change.

Internal improvements to product analytics and a small piece of preparation work for an upcoming payment-method addition. No user-visible changes in this release.

• ImprovedImproved internal analytics so the team can better understand how new visitors interact with the site and where the experience falls short. Existing privacy controls (consent banner, region-based opt-out) apply unchanged.
• ImprovedGroundwork for an additional payment method at checkout. The new option will appear automatically once the rollout is complete; no action needed on your side.

When dozens of your monitors fail at once because a shared origin server has a brief outage, you now get exactly one server-level alert naming the affected server, the count of impacted sites, and a sample of which ones. The individual monitor incidents are nested under that one parent incident on your dashboard, so the audit trail stays complete. A 30-second deferral on linked-monitor notifications gives the system time to recognize a cluster pattern before any noise reaches your channels. The full-feature trial offer also no longer auto-starts the moment you pick the Free plan.

• ImprovedServer-level alert aggregation. If your monitors are linked to a Server, and many of those monitors fail together within a short window, you receive ONE alert for that server. Per-monitor incidents are nested under one parent server incident on your dashboard, so you can see the full picture in one place. Recovery works the same way: one server-level recovery message instead of dozens.
• ImprovedWhen the system has direct evidence of server state (via the optional server agent script), the alert message says clearly whether the server itself appears offline, OR whether the server is responding but the hosted sites are unreachable (a network or app-level issue). Two different problem shapes, two different message framings.
• ImprovedPer-channel notification cap as a final-line safety net. Each integration channel (Discord, Slack, WhatsApp, email, push) has a per-organization cap on how many notifications can fire in 60 seconds. If the cap is exceeded, further notifications on that channel are held and a single summary message is delivered telling you how many were suppressed. Keeps Discord and Slack channels usable during any kind of widespread incident.
• ImprovedPicking the Free plan no longer starts a 7-day trial clock the moment you sign up. You can take your time setting up your first monitor without watching a trial expire in the background. The full-feature trial offer will surface later, at the moment you actually need it (when you hit a Free-tier limit or click on a paid feature).

• FixWhen you invite a teammate and they sign up, they no longer get a phantom personal workspace before joining yours. They land directly in your organization.
• FixFree plan details on the website-monitoring overview page now match what you actually get on the Free plan: three monitors, HTTP and SSL checks, email and push alerts, EU and US regions, no time limit.
• ImprovedComparison pages updated for accuracy. We tightened the wedge against other tools to focus on what is genuinely different (visual regression, content change detection, real-browser checks) rather than feature gaps that no longer exist.

Detection of registrar parking pages and hosting-suspended landings, introduced one release earlier, now applies to checks from every region rather than only some. Multi-region checks classify the same situation the same way regardless of which region observes it.

• FixParking-page and hosting-suspended classification now applies uniformly across all check regions. Previously the new detection only ran in some regions, which could produce a partial-outage state on a domain that was actually parked everywhere.

When your domain stops pointing at your hosting (lapsed domain billing, transferred away, parked at the registrar), Visual Sentinel now classifies the situation correctly at the uptime layer instead of firing a confusing visual-difference alert with a misleading "all metrics normal" diagnosis.

• FixWhen a domain redirects to a parking, suspended, expired, or maintenance landing page, the uptime check now flags the monitor as down and the alert clearly states the cause (e.g. "Domain redirects to a parking/suspended page"). Previously these situations could fire a misleading visual-difference alert with a server-health diagnosis that did not reflect the real cause.
• ImprovedVisual checks no longer run when the underlying page has been replaced by a registrar parking page or a hosting-suspended notice. This avoids comparing the wrong page against your baseline and saves resources for the actual sites you are monitoring.

Two mobile fixes from real reports. The top navigation now stays accessible while scrolling on iPhone Safari, and the support chat icon is calmer and smaller on small phones.

• FixTop navigation now stays in place while scrolling on iPhone Safari. The hamburger menu is reachable from anywhere on the page, not just the top.
• ImprovedSupport chat icon on small phones is now smaller, with a calmer animation. The "Hi!" greeting pops up much less often, and reduced-motion preferences are now respected fully.

Audit log entries for password and admin actions now record reliably. Refreshed brand mark and favicon for sharper rendering on light and dark themes.

• FixSuccessful password resets, password changes, and admin actions now record in the account audit log reliably.
• FixForm submissions opened just before a release no longer fail when submitted after the release completes.
• ImprovedRefreshed brand mark and favicon. The owl logo now adapts to light and dark themes consistently across the marketing site, dashboard, and browser tabs.

SSL certificate expiry alerts are now tuned for less noise on healthy auto-renewing certificates while preserving early warning for actual renewal failures. Plan, channel, and feature descriptions across the marketing site, blog posts, and AI-agent documentation now exactly match what each tier offers.

• ImprovedSSL certificate expiry alerts now default to 13, 3, and 1 day before expiry. Optional 20 and 6 day tiers remain available per monitor for users who want extra cushion. Existing monitors keep whatever tiers they already had selected.
• ImprovedPlan, channel, layer-coverage, and retention descriptions across the homepage, pricing, blog, and AI-agent documentation now exactly reflect what each tier includes. Several legacy claims that overstated the Free tier or understated the Business tier have been corrected.
• ImprovedWeb Push is now visible alongside the other alert channels on the homepage so its availability is obvious at a glance. Web Push is enabled per browser with a single click and lands directly on the operator device.
• ImprovedAI-agent documentation files (/llms.txt and /llms-full.txt) restructured with verified customer incident write-ups, a canonical positioning section, and per-section indexes at /tools/llms.txt, /compare/llms.txt, and /blog/llms.txt.
• FixMonthly product update email subject now correctly references the month being summarized (the previous month) rather than the month it sends in.
• ImprovedBeta environment network reliability improved (transport and routing changes on the beta host).

Added two new comparison pages, two new long-form blog posts, and deeper coverage on the visual monitoring and server monitoring pages. Homepage and pricing copy now describe each plan in plainer language so you can pick the right tier faster.

• NewTwo new comparison pages added under /compare.
• NewTwo new long-form blog posts added under /blog.
• ImprovedHomepage now describes the typical visitor types in clearer language so each reader recognises themselves earlier.
• ImprovedPricing tier descriptions are now plainer about who each plan fits.
• ImprovedVisual Monitoring page expanded with additional context on continuous change detection.
• ImprovedServer Monitoring page expanded with additional context for common server stacks.

Refreshed the public documentation file at /llms.txt with the latest pages and product surface so it stays a faithful summary of what is on the site.

• ImprovedPublic documentation file refreshed.

Updated the top navigation and footer to expose recently added pages so they are one click from anywhere on the site.

• ImprovedTop navigation now includes a direct link to the MCP server install page on both desktop and mobile menus.
• ImprovedFooter expanded with links to recently added pages.
• ImprovedAPI documentation page links to the MCP install page and the public OpenAPI repository.

Added a new side-by-side comparison page and a new /website-monitoring overview that walks through all six monitoring layers and how a check runs end to end.

• NewNew comparison page added under /compare.
• NewNew /website-monitoring overview page covering the six monitoring layers and how a check runs end to end.

Visual Sentinel is now installable in Claude Desktop, Claude Code, Cline, Cursor, and Continue via the public @visualsentinel/mcp-server package on npm. Ask your AI assistant what monitors are down, triage an incident, run a free SSL or DNS check, or onboard a new monitor by description, all from inside the conversation.

• NewVisual Sentinel is now installable in Claude Desktop, Claude Code, Cline, Cursor, and Continue. Install snippets are at /mcp.
• NewSixteen tools available, including five that work without an account.
• NewOptional Claude Code plugin bundle with task-focused skills and slash commands.