Auto Healing Agent for Reliable Test Automation

They act at different moments. Auto-Heal works on the locator during the run, re-resolving a lookup that failed so execution continues. Adaptive Heal works on the step: when an objective cannot be replayed at all, it re-authors that objective and every one after it, then saves the result as a new test version. Because Adaptive Heal changes the test rather than one lookup, it is held for approval, while Auto-Heal simply carries on.

No. Auto-approve changes ships off, so a test that Adaptive Heal or Dynamic Test rewrote is stored in Version History as a draft and becomes current only once someone approves it. If you do switch auto-approve on, the version is marked as approved that way, so the record still shows whether a person reviewed it. Adaptive Heal and Dynamic Test also need a test case authored with the New Experience engine, which is currently available on Desktop Web browsers only.

Every heal is recorded. Smart Heal logs the original and healed locators with screenshots, and a step that Adaptive Heal or Dynamic Test re-authored appears in Version History as its own version, so you can compare it against the one before and roll back. For a fuller picture of a run, Evidence Reporting covers coverage details and the business use cases behind the tests, and it applies where every test case in the run was authored with the New Experience engine and the run executes on Chrome. A single Classic test case, or any instance on another browser, makes the run fall back to the standard report.

Auto-healing works best with locator changes caused by UI updates. It cannot recover from WebDriver initialization errors, network failures, or complete page redesigns. All healing actions are logged so your team can review what was fixed and identify recurring issues.