Changelog

Find an element gains a Vision mode: describe what you want in plain language and the screen is read for you, instead of hunting through the hierarchy. The rest of this release is stabilisation, and a dense one: faster execution loading, a crash on load fixed, the password reset flow repaired end to end, and a long pass on the recorder, the Realtime views and the test editor.

👁️ New: Vision mode in Find an element

Open Find an element and switch to Vision: describe the element you are after, and the mode reads the screen to locate it for you. It picks up where the hierarchy stops being readable, on a canvas, a custom rendering, or a screen whose tree says nothing useful.

📬 Improved: sender filters in the Notifier

A sender can now be authorised by pattern, *@yourdomain.com covering every address of a domain, and the filters are split into numbered variables so a long list stays manageable.

🐛 Fixes

Stability and performance

• The app no longer fails to load with a crypto.randomUUID error

• Execution pages load much faster

• One flaky Android device no longer takes its whole agent down, iOS reboot commands are spaced out and stop once the device leaves the list, and a resource registration is no longer cut short by whatever triggered it

Password reset

• A reset link works even when the browser is still signed in

• A malformed token answers with a clear error instead of a server error, and blank fields are treated as no link at all

• The password policy no longer returns its verdicts without a valid token

Recorder and element picking

• The "Use recorder" button works on an URL without HTTPS, and on Microsoft Edge

• The smartphone XML modal respects the screenshot ratio and places elements correctly on the devices where it was off

• Find an element uses the right coordinate scale for iOS on execution results

Test editor

• Test V2 no longer renders oversized

• The selector field is no longer truncated on the steps that carry an operator

• The "after" and "after each" hooks run when a test fails

Realtime

• The browser back button returns to the view you had chosen, not the default one

• The view editing options work again, and the details page no longer scrolls twice

• The By product pages of the menu lead where they say

Alerts and incidents

• The SMS alert regression is fixed

• failure_reason is carried in the payload when the test result has one

• The incidents dropdown loses its stray red dot, and clearing a filter clears its value too

Elsewhere

• International MSISDN flags are no longer truncated on the SIM details

• The resource usage status is consistent again

Two more destinations for your alerts: a Kapptivate incident can now open a ticket in ServiceNow or an issue in Linear, next to Slack, Teams and Jira. The test editor gains multi step selection, copy and paste between tests and a real save shortcut.

🔌 New: ServiceNow and Linear in the Integration Center

Two new connectors. An alert, warning or critical, and a robot disconnected notification can now open a ticket in ServiceNow or an issue in Linear, alongside Slack, Teams and Jira. Connect the tool once per workspace in Administration → Integrations, then pick it on your alerts.

⌨️ New: keyboard shortcuts in the test editor

Select several steps at once. Shift and click selects a contiguous range, Cmd or Ctrl and click adds or removes a single step from the selection.

Copy and paste steps, with Cmd or Ctrl plus C and V. The paste lands under the selected step, works on a multiple selection, and works from one test to another.

Delete keeps your place. After deleting, the step below is selected, or the one above when you deleted the last one.

Cmd or Ctrl plus S saves the test, without the browser's own save dialog getting in the way.

No shortcut fires while your focus is inside a field, so typing in a name, a selector or a value stays safe.

✨ Other improvements

• Steps are numbered in execution results, with the same numbers as in the editor, so a failure can be named out loud and found on both sides

• A button indents the JSON body of an API Call step, instead of one unreadable line

• The What's new modal follows the language you chose, so the French release notes no longer open in English

• "Operator" becomes "workspace" on the last screens where the old word survived

• The whole selection column of a table toggles the checkbox, instead of the small square alone

• A step acting inside an iframe now waits up to 10 seconds for it, instead of 2, so a payment form that loads late no longer fails the step

• The BrowserStack provider is hidden for workspaces that do not have the option

🐛 Fixes

Tests and recorder

• Starting the v2 smartphone recorder no longer starts the v1 recorder in another tab and records twice

• The Get XML button of the Live session opens Find an element, like the recorder, instead of the old XML viewer

• The App source artifact of an execution result is no longer empty

• Copying from a step and pasting into the phone no longer leaves the focus behind

• The output variable of a Get sms code step can be renamed, and its placeholder is readable again

• Element descriptions containing line breaks no longer break their row

Executions

• Opening an execution directly by its URL no longer shows an error page on the first load

• The Screenshots tab no longer shows an empty tile for steps that have none

• The Export split button gets its separator back

Monitors and campaigns

• A monitor's start time can be set again

• Monitor names are readable on workspaces with long names

• Campaign monitors get their correct layout, their test count tag and their run mode selector back

Alerts

• An alert now opens an incident for tests whose name contains a + or an apostrophe

• An alert created before the Integration Center switches to the connector when you pick a channel, instead of keeping its old webhook

Custom step library

• Steps linked to a library script stay linked after a run, instead of turning back into plain custom steps

Realtime

• The status tag appears on every card, not only on some of them

Agents and devices

• A new iPhone or iPad no longer stays stuck in Configuration

• Live sessions and v2 web tests honour the DNS of the custom capabilities again

• The automatic reboot of a robot happens at the configured hour

• A SIM in 5G standalone is seen as registered, and a VoLTE call is no longer dialled before the modem is ready

• Five defects fixed on the Agents page: SIM badge, empty state, default tab, a translation and the Status filter

Visual assertion

• A visual assertion reached more than ten minutes into a test no longer fails on an expired token, and says what happened when it does

• A response the service cannot read is reported as a technical error instead of becoming a negative verdict

• A deployment no longer cuts a check in flight

Elsewhere

• The date picker of the Excel report is no longer hidden behind the page

• A series of UI corrections: breadcrumb, time picker, selects of API steps, alignment of the Live session history filters, disabled danger buttons and device modal spacing

KTM 8.95

No big feature announcement this time. We spent the last two weeks on the details you run into every day: steps that wait properly, fewer dead ends, clearer screens and a long list of fixes.

⏱️ Improved: Mail and SMS code steps wait the way you want

Mail steps and SMS code steps now let you control how long to wait for a message and how often to check for it.

• Timeout: How long the step waits. Mail steps now wait up to 60 seconds by default (was 20), SMS code up to 2 minutes.

• Interval: How often the step checks. Every 3 seconds for mail, every 5 seconds for SMS, with a 1-second minimum.

• No more guessing: When several messages match, the step fails right away instead of picking one randomly.

Both settings accept global variables.

📈 Improved: Create a campaign monitor from anywhere

Creating a monitor for a test campaign no longer means going through the Monitors page.

A Create monitor action is now available from:

• The row menu in the test campaigns list

• The selection banner

• The campaign page header

The monitor form opens with your campaign already selected.

Also new:

• Campaign monitors now offer the execution mode option, like single-test monitors

• Monitors whose last run ended in a technical error no longer show as successful

🔔 New: Robot availability alerts in Slack, Teams or Jira

Following the Integration Center launch, robot availability notifications can now be sent to a connected Slack, Microsoft Teams or Jira integration.

Jira tickets now include which incident triggered them, including when a robot disconnects.

Alert emails are back to the Kapptivate design instead of a plain title and bullet list.

🧭 Improved: Real 404 pages, and fewer dead ends

• 404 pages: A link to a test, monitor, execution, campaign, reusable component, device or agent that no longer exists now shows a proper 404 page with a way back to its list, instead of loading forever.

• Reusable components in use: The delete dialog now lets you remove the component from tests that use it, then delete or keep it.

• Variants in use: A variant used by tests, monitors or campaigns can no longer be deleted. Tests pointing at a deleted variant tell you clearly.

• Smarter run forms: No longer ask for a value for a variable the test writes itself.

• Safer saving: An edit made right after clicking Save is no longer lost. Renaming a test from the breadcrumb saves immediately.

🔐 New: Session expiration in security policies

Admins can now make sessions expire after a delay of their choice, from 1 hour to 365 days.

The option is off by default. Turning it on doesn't log anyone out, and API keys are not affected.

📱 Smartphone Improvements

• Better input field detection: The tap on a field is kept before typing, so the test focuses the field first

• Correct capabilities picker: Smartphone V2 tests now suggest Android or iOS capabilities instead of web ones

• SIM preservation: Tests created from Live Session history keep the SIM of each step

• Get XML upgraded: The Get XML button in Live Session now opens the new Find an element window

✨ Small things

• Menus and selects can be used with the keyboard, and selects stay fast on long lists

• Press keys step shows its keys in a field, like every other step

• What's new window shows release notes in your chosen language

• Your browser no longer offers to translate a French interface into French

• Bulk monitor creation window explains why creation is unavailable

• Admin user list shows when each user was last active

• AI features are now available on on-premise installations

🐛 Fixes

• Rebooting a robot works again

• Robot's Event logs tab lists its events instead of staying empty

• Queued executions show the device they will run on

• Clicking a long execution name in the list opens it correctly

• Usage tag on device/agent pages matches the weekly figures below

• Long device names no longer hide the phone number in the device field

• Enable AI checkbox is read-only when other test settings are

• Visual fixes on step chevrons, multi-line tooltips, failed steps inside step groups, device selection window and empty states