Changelog
What I've shipped, newest first, dated the day it went live. Follow along with the Atom feed.
November 2026
Short videos of the product at work
I recorded short screen recordings of StatusPulse doing its job, from the Setup Agent turning a domain into probes to an incident from first failure to post-mortem and an AI assistant running a status page over MCP. Each one sits on the page it explains, such as the AI agents, status pages and on-call, and five short clips are in Help. They have no sound and nothing downloads until you press play.
Latency charts no longer dip below zero
Next to a single slow check, the latency line on the dashboard and on public status pages curved below 0 ms and above the real peak. It now follows the measurements without overshooting. Deleting a probe also brings you back to its own status page instead of your first one.
Maintenance and incident toggles show up again
In the Schedule maintenance window, "Auto-open at start time" and "Exclude from SLA uptime" had no visible switch and their labels wrapped a word per line; the same happened to "Public update" and the maintenance settings on an incident's page. They're proper switches now, and the affected-components list no longer pushes long probe names under their switches.
Report names and deleting from page two
A generated report's page was titled "Last7Days report"; it now reads "Last 7 days report", as the list and the schedule already do. Deleting a report from the second page of the history now brings you back to that page instead of the first. On the API tokens page, the quickstart no longer carries a leftover internal note.
Heartbeat URLs from the Probe Wizard, and Rotate
A heartbeat added through the Probe Wizard went straight back to the probe list without ever showing its URL, and the Rotate button on the probe's Settings tab saved the probe instead of issuing a new URL. The wizard now opens the new heartbeat with its URL ready to copy, and Rotate asks for confirmation and shows the new URL. If you have a heartbeat that never received a beat, open its Settings tab and click Rotate.
On-call rotations and escalation steps can be built again
Adding a layer to an on-call schedule, or a step to an escalation policy, failed with "The Name field is required." even with every field filled in. Both now save, so you can set up who gets paged, in what order and after how long. Until you add steps, pages keep going to your workspace's Owners and Admins as before. The steps table also reads in plain words now ("Specific member", "Email, SMS").
Error messages stay on screen long enough to read
When something failed, the error message vanished after three and a half seconds, so an AI branding analysis that couldn't reach your website looked as if nothing had happened. Errors and warnings now stay for ten seconds (click to close them sooner), and when a site refuses our servers the message points you to the From screenshot tab, which works for any site.
On-call history and paging fixes
The on-call overview and history now list only your own workspace's paging events, and an incident whose status page no longer exists no longer pages anyone every 30 minutes. The person list under Watchers and the report history now move past their first page.
The copilot's drafted update keeps its paragraphs
"Use the drafted update" on an incident put the copilot's text in the update box as one run-together paragraph. It now arrives with its line breaks, ready to edit and post. On the dashboard, long probe names in the page cards now end in "…" instead of being cut on both sides. And the incident link your AI assistant gets back from the agent API now opens the incident instead of a "page not found".
The AI, front and centre
The home page now leads with what the AI in StatusPulse does across an incident: the Setup Agent sets up your probes, forecasts flag probes at risk, the copilot finds the probable cause, AI drafts your updates and post-mortems, reports come with a plain-language summary, and your AI assistant can run it all through MCP. There's also a new FAQ answer on what the AI does with your data.
Deleting a failing probe closes its automatic incident
If you deleted a probe while it was down, the incident StatusPulse had opened for it stayed open, because a deleted probe never checks in again to recover. Deleting it now resolves that incident, with an update saying the component is no longer monitored; if other probes on the incident are still failing, it waits for them as before.
Run your status page from your AI assistant
It's now easier to find out that Claude Code, Cursor or VS Code can run StatusPulse for you through its MCP server: put new services on your status page, tell you what's down, and open, update and resolve incidents with your approval. There's a new section on the home page, a Connect AI assistant button on the dashboard, and a guide to setting it up.
statuspulse.ai/status opens our live status page
The /status address on this site showed a placeholder that always read "All systems operational". It now takes you to our real status page, which runs on StatusPulse and shows the live state of the platform.
No product analytics once you sign in
I now run product analytics only on the public website and the sign-up form: nothing is tracked once you sign in, and visits are never linked to your account. The privacy policy and the GDPR page say so.
A home page that says what StatusPulse does
The home page now says in one line what StatusPulse is, and has a one-minute tour of adding a probe next to a link to our own live status page, which now also watches the dashboard, the status pages, the probe network, our ping worker and DNS.
Connect your AI assistant with one prompt
Under Pages → API tokens, Create setup prompt gives you a ready-made prompt: paste it into Claude Code, Cursor, VS Code or another assistant that can add MCP servers, and it connects itself to your workspace, checks the connection and offers to set up your monitoring. The prompt carries a new agent token that you can revoke at any time.
Let your AI coding agent run your monitoring
Owners and Admins can now issue an agent token and hand StatusPulse to Claude Code, Cursor or any MCP client: it can create status pages and probes, run the Setup Agent on your domain, tell you what's down and open or update incidents, within your plan's limits. The instructions your agent needs are at statuspulse.ai/agent.
Incident copilot: the probable cause, a draft update and who to page
On Pro and Business, every incident now has a copilot: it reads all the probes on your status page and the timeline around the incident, then suggests up to three likely causes with the evidence for each, drafts your next update and says who on call to page first. On incidents StatusPulse opens automatically it runs as soon as you open them; nothing is posted or paged without you.
Setup Agent: give it your domain, get your probes
Type your domain and the new Setup Agent looks at its DNS, certificates, subdomains and health endpoints, then proposes a status page with the probes worth having — website, API, app, certificates, domain and mail — for you to review before anything is created. It's on every plan, including Free.
Switch AI assistance off for your workspace
Owners and Admins can now switch every AI feature off for their workspace under Account → Data use; from then on nothing from it is sent to our AI provider. The privacy policy explains what each AI feature sends.
October 2026
Old forecast models are now deleted
When a newer anomaly model replaces one, I now delete the old model and its file within 30 days, instead of keeping every version forever. If your workspace switches off model training, the models trained on its data are gone within 30 days of being replaced; the details are in the privacy policy.
Signing in after closing a workspace
If you close your workspace and sign in again during the 30 days, I no longer quietly hand you a new, empty one: you see when it will be deleted and how to have it reopened, and you can start a new workspace yourself if that's what you want.
Your audit log shows changes, not probe checks
Ping checks, heartbeats and the WebSocket, gRPC and database timings were adding a row to your audit log every time they ran. Now only changes people make are recorded.
More control over a custom domain after a downgrade
On a plan without custom domains you can now deactivate or remove the domain yourself, and once its 14 days are over our emails link straight to your page on status.statuspulse.ai. Closing a workspace also frees its single sign-on domain at once.
Business probes run from both locations every interval
On Business, a check from West Europe could make the US location skip its turn, so some probes were rarely checked from the US. Each location now keeps the probe's interval on its own, and checks in the same round run side by side, so one slow target no longer delays the rest. Probes
Close your workspace from the Account page
Owners can now close a workspace themselves: everything stops at once, I can reopen it for you for 30 days, and after that everything in it is deleted for good. Cancelling a paid plan still only moves you to Free and deletes nothing. Privacy
Custom domains and post-mortems after a downgrade
If you move to a plan without custom domains, your domain now keeps serving your status page for 14 days and then sends visitors to the same page on status.statuspulse.ai; upgrade and it's back with nothing to redo. Post-mortems you already published stay public, and only writing new ones needs Pro or Business.
Editing a post-mortem keeps your changes
Saving or publishing a post-mortem that already existed could put the previous text back instead of your edits. Now what you type is what gets saved and published (post-mortems).
Opt out of anomaly model training
Any workspace can now keep its check history out of the shared forecast models with one switch under Account → Data use, on every plan. The next hourly training run leaves it out, and your own forecasts keep working. Anomaly detection
Clearer feature pages and fairer comparisons
I rewrote the feature pages from the product's own code: real limits per plan, examples and what happens on a downgrade, with small features folded into the page they belong to. The comparisons now link each competitor's own pricing and docs, and I corrected the claims that had gone out of date.
Deleting a status page no longer moves its probes
Deleting a status page used to move its probes onto another of your pages. Now the page and its probes stop together and wait 30 days under "Pending deletion" in the page picker, where you can restore them or delete them for good.
30-second checks on Business
Business probes now really run every 30 seconds: HTTP, Ping, TCP, DNS, WebSocket, gRPC and database checks are scheduled twice a minute, and the interval is picked from a list that starts at your plan's fastest. Pricing
Probes run exactly on their interval
A probe checked a few seconds into the scheduler's minute was skipped at the next one, so 1-minute probes ran about every 2 minutes. Every probe now runs on its interval.
Paid features follow your current plan
On-call paging, scheduled reports, Slack, Teams, SMS and webhook notifications, the custom Reply-To, white-label footers and public post-mortems now check your plan each time they run, and the per-page subscriber limit applies to the Subscribe button too. Moving to a smaller plan never deletes anything: upgrade again and everything comes back as you left it. Pricing
Plan features are checked wherever you set them up
Invites, manual and internal incidents, incident metrics, CSV export, SLO impact, component groups and API tokens now follow your plan when you set them up, not just in the menus. Nothing you already have is removed when you change plan: a saved custom domain, Slack or Teams webhook and your report schedules are kept, and you can still see and delete the schedules. Pricing
Anomaly alerts, one switch per probe
Business HTTP, Ping, TCP and Heartbeat probes now have an Anomaly alerts switch on their Forecast tab. Turn it on and everyone watching with Forecast enabled hears about a likely outage before it starts, by email, SMS, Slack or Teams. Each warning goes out once and only counts checks made after you switch it on. Anomaly detection
Probe history follows your plan
Raw check results are now kept for your plan's history period (never less than 35 days), and every day is first summarised into a daily uptime and latency record that is kept for good. Long-range uptime and annual reports keep working, and the database stays fast. Pricing
Heartbeat probes no longer alert on runs that are on time
A heartbeat that arrives inside its grace period now counts as on time, so cron jobs stop sending an alert every run, a ping a moment before the cron time counts for that run, and the first run after a missed one clears the alert straight away. You can now also set the grace period in cron mode. Cron heartbeat guide
The GDPR page on phones
The sub-processor table on the GDPR page made the page wider than a phone screen. The table now scrolls sideways on its own and the page fits. GDPR
Slack, Teams and SMS alerts no longer need an email watcher
A probe with nobody watching it by email skipped its Slack, Teams and SMS alerts too. Those channels now get the alert either way.
Escalating to a watcher group pages people
An escalation step pointed at a watcher group paged no one. It now pages the people in that group who are members of your team.
Copying a probe keeps all its settings
Copying a probe to another page used to carry only the basic fields. It now copies every setting, credentials included, and a copied Heartbeat probe shows its new URL straight away.
Probe timeouts
DNS probes now respect their timeout and always ask the resolver you chose directly. The HTTP probe timeout now goes up to 30 seconds, the limit the check always enforced, and a WebSocket timeout now says how long the handshake waited.
The site now says where your data lives
Some pages said you could pick US or EU hosting. That choice doesn't exist yet: all your data is stored in the EU (Azure West Europe), and Business plans also run checks from a US location. Every page, including Privacy and GDPR, now says exactly that. GDPR
Reports, maintenance updates and subscriber emails go out once
StatusPulse runs its background jobs in two regions, and a few of them could each send the same thing: a scheduled PDF report, the "maintenance started / ended" update with its notifications, or a subscriber email. Each now goes out exactly once.
The Help matches the product
I went through every Help page against the product. Field names, defaults, limits and plan tiers now match what you see on screen, and the Help no longer describes options that don't exist yet, such as paging by Slack or voice.
On-call escalation pages now reach the person on call
Escalation pages for stale incidents weren't being sent: the background job that pages people couldn't see the schedule or the person it was meant to page. They go out now, and a step that pages by email and SMS escalates to the next step once, not twice.
The blog moves to statuspulse.ai
The blog now lives at statuspulse.ai/blog: practical answers on uptime, incidents and status pages, with an RSS feed for your reader. Read the blog
Export your whole workspace
Owners and Admins can now download every probe, status page, subscriber and incident as one JSON file from the Account page, on every plan. Credentials stay out of the file. Leaving should be as easy as joining. What if StatusPulse shuts down?
14-day money-back guarantee and a pricing page
Every paid plan, monthly or annual, now has a 14-day money-back guarantee on your first payment. The new pricing page compares every plan feature by feature, straight from the limits the product enforces. Pricing
A shorter sign-up form
Signing up now takes a name, an email and a password, with a live checklist for the password rules. Your email is filled in for you on the first sign-in. Start free
The homepage on phones
On a phone the homepage was wider than the screen, the headline was cut off and the menu showed only Login. It now fits small screens, and a menu button opens every link.
September 2026
Hide the Subscribe button on a status page
You can now switch off the Subscribe button on any status page, on every plan. When it is off, the button and the sign-up form are gone entirely, which suits teams that keep their subscriber list elsewhere. Status pages
Second round of security hardening
I finished a second full security review of the codebase and shipped the resulting fixes across the dashboard, the public status pages and the APIs. Custom webhook headers are now encrypted at rest too, and there is nothing you need to change. Security
July 2026
Page Wizard and Probe Wizard
The Page Wizard, in the Pages menu, takes a new account from an empty dashboard to a live probe in a few steps. The Probe Wizard adds any probe type your plan includes to an existing page, asking only for the required fields.
Sign-up could fail when a page address was already taken
Some new accounts could not complete their first sign-in because the address picked for their default status page was still held by a deleted page. Sign-up now chooses a free address instead, and you can rename it later.
A lighter, faster dashboard
The dashboard now reads 30-day uptime from daily summaries instead of scanning every raw check, and I removed a stream of background writes that competed with it for the database. Each visit now does far less work.
June 2026
Single sign-on with your own identity provider
Workspace owners can connect StatusPulse to their company's OpenID Connect identity provider after proving domain ownership with a DNS TXT record. Members then sign in with their work account, and newcomers can join automatically on first sign-in, within your seat limit.
Security hardening across the dashboard and public pages
I ran a full security review and shipped the fixes in three rounds, across the dashboard, the public status pages, probes and integrations. Slack and Teams webhook addresses are now encrypted at rest, and there is nothing you need to change. Security
Faster public status pages
Public status pages now fetch only the 48 hours of checks the heatmap shows, work out 30-day uptime in the database and sit behind a ten-second cache. Each visit costs a fraction of the work it used to.
May 2026
On-call schedules and escalation policies
I added on-call rotations and escalation policies. When a major or critical incident goes quiet, StatusPulse pages whoever is on call by email and SMS, with a one-click acknowledge link, and escalates if nobody responds. On-call rotation
Restrict your workspace to your company domain
You can now require every workspace member to use an email address on your company domain. Invitations can only be accepted from addresses on that domain. Domain lock
White-label status pages and a custom Reply-To
You can remove the "Powered by StatusPulse" footer from a status page and add a footer line of your own. Each page can also set a Reply-To address, so replies to alert emails reach your support inbox instead of ours. White-label
Public read API and audit log export
You can create API tokens and read your page's status, incidents and probes as JSON, with rate limits that grow with your plan. The same tokens also export your workspace audit trail, filtered by date, person, entity or action. Read API
Incident proposals from anomaly detection
When anomaly detection spots a sustained problem on a probe, StatusPulse now drafts a proposed incident in your dashboard. It stays private, with no public post and no notifications, until you promote it with one click or dismiss it. Anomaly proposals
Error-budget impact on every incident
Each affected component on an incident now shows its uptime during the incident, the estimated downtime and how much of its monthly error budget the incident used, based on the probe's SLA target. SLO impact
Internal incidents
You can mark an incident as internal to track it with your team without publishing it. It never appears on the status page and its updates stay private. Internal incidents
Incident metrics and CSV export
A new Metrics view shows mean time to resolve, time to first update, incident volume and the components involved most often, per status page over 30, 90 or 365 days. Any incident's timeline can also be exported to CSV. Incident metrics
Component groups
You can split a status page into sections such as Frontend, API or Database, so visitors read it as a map of your system instead of one long list of probes. Component groups
Trust score on public status pages
Status pages now show a service-quality panel with average time to resolve, time to first update, updates per incident and post-mortem rate over the last 90 days. It appears after three resolved incidents and stays hidden if you hide incidents. Trust score
Status badges and an embeddable widget
Every status page now has an SVG status badge for READMEs and docs, a shields.io-compatible endpoint and an embeddable widget. The widget supports oEmbed, so tools that unfurl links can show your live status. Status badge
Automatic incidents could fail to open
Incidents that StatusPulse opens on its own after consecutive failed checks could fail to save, so the incident never appeared and nobody was notified. They now open and notify as intended.
Outbound webhooks for PagerDuty, Opsgenie and Zapier
Each status page can send signed incident webhooks to tools such as PagerDuty, Opsgenie or Zapier, with your own request headers and automatic retries. Outbound webhooks
Self-service sign-up and billing
You can now sign up and pick a plan on your own, monthly or annually, with payment by card through Stripe. Upgrades, invoices, card changes and cancellations are all handled from your Account page.
Team invites and roles
Invite colleagues to your workspace as Owner, Admin, Editor or Viewer, and switch between workspaces from the user menu. The number of seats follows your plan. Team invites
Checks from Europe and the US
On plans that include multi-region checks, probes now run from both Europe and the US. Latency charts and PDF reports break response times down by location, so a regional slowdown no longer hides in the average. Multi-region checks
Approve subscribers before they get alerts
Visitors can ask to be notified from the Subscribe button on your status page. Each request waits in a Pending approvals queue until you accept or reject it, so nobody receives alerts you have not approved.
Status pages, reports and alerts in six languages
Status pages, PDF reports and alert emails now come in English, Portuguese, Spanish, French, Italian and German. Visitors can switch language on the page, and each subscriber receives alerts in the language they chose. Multi-language
Custom branding for status pages
On Pro and Business, each status page can carry your logo, six brand colors, a font and a text size, in light and dark themes. Paste your website address and AI suggests matching colors and a font for you. Custom branding
Database probe for PostgreSQL, MySQL, SQL Server and Redis
On Business, a probe can sign in to PostgreSQL, MySQL, SQL Server or Redis, confirm the server answers and optionally run a SELECT query of your own and check the result. Passwords are encrypted at rest. Database probe
WebSocket and gRPC Health probes
On Pro and Business, WebSocket probes check the upgrade handshake, the subprotocol and round-trip time over HTTP/1.1 or HTTP/2. gRPC Health probes call the standard health service, per service or by streaming Watch, with mTLS and HTTP/3 support. Probe types
PDF reports
Generate a PDF report for a status page covering availability, latency, incidents, maintenance and post-mortems, or have one emailed every week, month, quarter or year. An optional AI summary explains the period in plain language. PDF reports
April 2026
Incident manager with a public timeline
Open an incident, post updates as it moves from Investigating to Resolved, and mark which components are affected. Everything appears on your status page timeline, subscribers are notified, and each page gets RSS and Atom feeds. Incidents
Post-mortems and scheduled maintenance
Write a post-mortem in a sectioned editor, keep it as a draft and publish it next to the incident on your status page. Scheduled maintenance is announced in advance and can start and end on its own. Post-mortems
AI-drafted incident updates and post-mortems
StatusPulse can draft an incident update or a post-mortem outline for you from the incident's timeline. You review and edit the text before anything is published. AI drafts
Anomaly detection and forecasts
A model trained on probe history flags patterns that tend to come before a slowdown or an outage. HTTP, ping, TCP and heartbeat probes get a Forecast tab, and you can turn on early-warning alerts per probe. Anomaly detection
Email round-trip probe
On Pro and Business, an email probe sends a real message through your SMTP server and can confirm it arrives in a mailbox over IMAP or POP3, timing each step of the journey. Email probe
Heartbeat monitoring for cron jobs
Give a cron job or background worker its own heartbeat URL to call when it runs. If the call is late, or the job reports a failure, the probe goes down and your alerts fire. Cron heartbeats guide
Ping (ICMP) probes
You can now ping any host over ICMP, on IPv4 or IPv6, and track its round-trip latency like any other probe. Ping probe
TCP, DNS and domain expiration probes
I added three probe types: TCP port checks with optional TLS and banner matching, DNS record checks with assertions on the answer, and domain expiration tracking that warns you before a registration lapses. All probe types
SSL certificate monitoring
SSL probes track how many days your certificate has left and warn you before it expires. You can also show the expiry on your public status page. SSL probe
Slack, Microsoft Teams and SMS alerts
Connect a status page to Slack or Microsoft Teams with an incoming webhook to post alerts to a channel. People on your watcher list can also get SMS alerts, with country-aware phone numbers and a monthly allowance per plan. Slack integration
Latency and SLA history for every probe
Each probe now has latency and SLA history charts, on your dashboard and on the public status page. You and your visitors can see how a service behaved over time, not only right now.
Email alerts for people and groups
Add the people who should hear about outages, organize them into groups and choose which probes each one watches. Alerts arrive by email, and every message carries an unsubscribe link.
Custom domains for status pages
Serve a status page from your own domain, such as status.example.com. The dashboard lists the CNAME and TXT records to add, checks them for you and sets up HTTPS automatically. Custom domains
HTTP monitoring and public status pages
The first version of StatusPulse: HTTP probes check your endpoints every minute, and each status page shows a 48-hour heatmap, uptime and an SLA target per probe. Slow responses show as degraded, and pages come in light and dark themes. Status pages