Skip to main content

September 2026

Molly Exten
  • Updated

New | Enhancements | Changes | Bug fixes

New

Teams testing private environments over mabl Link can now upgrade to Link 3.0. Among the highlights:

  • Faster connections and better recovery from network interruptions
  • Company-scoped tunnels to manage cloud traffic across workspaces from one set of agents
  • Personal tunnels to test local builds with the full results from mabl cloud runs
  • Control over which users can reach your tunnels
  • Easier installation

Link 3.0 is turned on per account once your agents are upgraded. Learn more.

Hand off more test setup to the mabl agent

Three additions let you describe what you want and leave the assembly to mabl:

  • Generate a similar test - most of the time you duplicate a test because you want something close to it, not identical. The duplicate button on browser tests now offers the option to use the duplicated test as a baseline and tell the mabl agent how the new one should differ. Learn more.
  • Manage DataTables from your AI client - create and edit DataTables, associate them with tests, and run a single scenario, all from your coding agent with the mabl MCP. Learn more.
  • Seed your workspace with files - upload the files your tests depend on from the new file uploads page or the mabl CLI, so the agent can complete a file upload step without stopping to ask you for the file. Download a file to reproduce a failing upload with the exact file the test used. Learn more.

Create, rename, and delete workspaces with the mabl API

For teams that keep a workspace per product, team, or client, workspace setup becomes a scripted step in onboarding instead of a trip through the account dashboard. A new company-scoped Workspace lifecycle API key covers the three operations and nothing else, and a new workspace can receive its credit allocation at creation time. Learn more.

Reuse a signed-in session across a plan run

When every test in a plan starts by logging in, it increases the plan's run time. Now you can configure plans to persist the logged in state and skip ahead to validate the goals of the tests. Turn on Persist browser state after flow execution for a login flow, and the first test to run it stores the session for later stages of the same plan run. Tests that share the application and credentials pick up the saved session and skip the login. Learn more.

Enhancements

Add custom headers to your webhooks

If your webhook receiver sits behind an API gateway or a token-checking proxy, mabl's fixed request headers used to rule it out. Now a webhook can carry up to 20 custom HTTP headers, so an Authorization token or a routing header travels with every request. Learn more.

Get desktop notifications for your mabl agent sessions

Agent sessions run in the background, which makes them easy to forget. You can now set up the mabl Desktop App to notify you when a session needs your input, finishes, or fails, with a sound and a dock badge if you want them. Learn more.

Manage users and spending across your account

Two additions to the account dashboard for admins who look after many workspaces:

  • Add a user to many workspaces at once - from the Users tab, add someone to one workspace, several, or every workspace in the account, and update their role where they're already a member. Learn more.
  • Billing alerts at more thresholds - in addition to the 85% alert, email every account admin when credit usage crosses 50%, 75%, or 90%, so an overage is a conversation before it's a surprise. Learn more.

Accessibility checks now use axe-core 4.13

Accessibility checks run on the latest rules from Deque: fewer false positives for color contrast and target size, three new rules for unnamed tabs, summaries, and table header attributes, and a lower severity for pages that block zoom. Learn more.

Recover from transient failures with Runtime Recovery Refresh

When a step fails because a page rendered late or a request stalled, Runtime Recovery Refresh reloads the page and retries the step once before failing the test. A test that would have failed for a reason unrelated to your application keeps going, and the run output shows where a reload was used. If you're interested in turning on Runtime Recovery Refresh for your account, reach out to the mabl team. Learn more.

Spend less time targeting the correct element on the page

The Trainer has two new improvements for finding the correct element:

  • Draw a box in any browser step - visual find, which targets an element by its appearance, now works for enter text, select, create variable, and assert steps, not only clicks. Learn more.
  • Let mabl write the locator - in the custom find editor, select Pick an element to generate, click the element, and mabl writes a recommended locator with a live count of how many elements it matches. Learn more.

Run teardown steps only when a test fails

Teardowns run after every test, but some cleanup or diagnostics only make sense after a failure. Three new runtime variables resolve inside a teardown, including run.test_status and the failing step's message and number, so a conditional can gate steps on the outcome. Learn more.

Manage your workspace report schedules in one place

The workspace report is configured per user per workspace, which used to mean visiting each workspace to change it. User settings now lists every workspace's schedule in one table, lets you update several at once, and sets a default for workspaces you join later. Learn more.

Changes

Link 3.0 replaces the legacy Link service, which shuts down on March 15, 2027. Before then, upgrade your Link Agents to version 3.0 or later, allow the new endpoints through your firewall, and contact mabl to move your account over. Tests that don't use mabl Link are not affected. Learn more.

Azure Pipelines task requires agent 4.248.0 or newer

The mabl Azure Pipelines task now runs on Node.js 20 or 24, ahead of Microsoft removing Node.js 16 from new agents in November 2026. Pipelines that reference mabl-azure-devops-extension@1 pick up the new version automatically. Microsoft-hosted agents already qualify. If you run self-hosted agents older than 4.248.0, update them, or the task stops before any mabl tests run. Learn more.

Bug fixes

General

  • Fixed an issue that returned an error when editing a test's parameters in a stage of a link crawler plan.
  • Fixed an issue where the plan Tests tab no longer showed the branch indicator for tests that exist only on an unmerged branch, so it wasn't clear which tests the run would skip.
  • Fixed an issue where an interrupted branch merge left the branch unable to retry and showed intact tests and flows as deleted.
  • Fixed an issue where certain users behind a corporate proxy couldn't log in to mabl and saw an unexpected error message.
  • Fixed an issue where Download CSV on the Tests, Results, and Plans tables exported every row instead of only the selected ones.
  • Fixed an issue where the account engagement report and its CSV export listed the same user more than once.
  • Fixed an issue where dynamic scenario selection on larger DataTables returned a different number of matching scenarios each time and sometimes failed with an error.
  • Fixed an issue where the scenario dropdown in the plan Test variables panel was hidden behind the Done button.
  • Fixed an issue where opening the Credentials dropdown for the first time on the new browser test form crashed the page.
  • Resolved an issue where the version history for certain flows failed to load.

Test authoring

  • Fixed an issue where generated step descriptions for assertions with a configured find read like wait until steps.
  • Fixed issues with dropdowns in the Trainer that rendered behind other elements in the API step editor or required scrolling in the find elements step instead of expanding.
  • Fixed an issue where the Current value preview in the Trainer didn't resolve a variable used as an array index.
  • Ensured that the Create flow button explains why it's unavailable when the selected steps are inside an existing flow or step group.
  • Fixed an issue where the CSS view in the Trainer's find element panel showed only the color attribute.
  • Fixed an issue where saving a file upload step that used a file Downloaded in this test failed unless the download step had been replayed first.
  • Fixed an issue where Cancel and Save in the Trainer's assertion edit panel stopped responding after reselecting an element.
  • Fixed issues that prevented adding assertions in the Trainer for certain elements, including repeated elements, tables inside dialogs, and elements in iOS mobile tests.
  • Fixed an issue that caused mabl agent sessions to fail partway through editing a test.
  • Fixed an issue where the mabl agent could save test edits that the step editor would have rejected as invalid.
  • Fixed an issue where the flow list in the Trainer's Import flow dialog took several seconds to load on first open.
  • Fixed an issue where flows imported into a test saved without their steps.
  • Ensured that innerText is available as an assertion attribute even when the element has no text content.
  • Fixed an issue where certain tests appeared empty after a related test that shared one of their flows was deleted.
  • Fixed an issue where flow parameter values on a test could be cleared without any edit to the test or the flow.
  • Fixed an issue where flow parameters appeared in a different order in the Trainer than on the flow details page.
  • Fixed an issue where pausing playback in the Trainer didn't interrupt navigation, download, or tab-await steps.

Test execution

  • Fixed an issue in Desktop App 2.68.4 that prevented local iOS mobile test runs from starting.
  • Fixed an issue where a while loop nested inside an if condition ran during cloud runs even when the condition evaluated to false.
  • Fixed an issue where Wait until element is not present steps ignored their configured timeout and failed after a few seconds.

Test output

  • Fixed an issue where the Show button next to redacted header values in cloud run network logs didn't reveal the value.
  • Fixed an issue where the pass and fail counts on the dashboard didn't match the Results page for users outside the UTC time zone.
  • Fixed an issue where columns added to the Results table couldn't be sorted or filtered and were missing from the CSV export.
  • Fixed an issue where failure analysis kept reporting an earlier cause on later runs of the same test and didn't rewrite its synopsis when corrected.
  • Fixed an issue where failure analysis stayed in the analyzing state and the Categorize failure option didn't appear for certain failed runs.
  • Fixed an issue where a step's description in cloud run output differed from the description shown in the Trainer.

mabl API

  • Fixed an issue where JavaScript snippet steps created through the mabl API or the mabl MCP were saved in a form the Trainer and test runner couldn't read.

Subscribe to our monthly newsletter

Interested in getting monthly updates sent directly to your inbox? Click here to subscribe to the monthly product newsletter.