| Term | Meaning |
|---|
| AI engagement | One decision made by the AI. The unit your license is measured in. |
| Ambiguity warning | A non-blocking note that a step could be read more than one way. |
| Assertion Lock | The rule that tuning may rewrite a step’s action but never its expected result. |
| Autonomous Tuning | Detection and correction of test wording that no longer matches the application. |
| Browser grid | Real Chrome and Microsoft Edge browsers on BrowserStack that a cloud run can use instead of the built-in browser. |
| Child collection | A dataset sheet whose rows do not line up with iterations — a list attached to one record. |
| Dataset | An uploaded spreadsheet supplying values to a test case. |
| Evidence report | The printable record of a finished run, from Reports or the run page. |
| Expected result | The acceptance criterion evaluated at the end of a test case (or after a step). |
| Failure triage | Automatic classification of a failure as app, test, environment or data. |
| Iteration | One run of a test case against one dataset row. |
| Local agent | Software running on your machine that executes tests against internal applications. |
| Memory recipe | A remembered way of locating an element, reused on later runs. |
| Not verified | A step that ran and passed without its expected result being confirmed. |
| Regression | A test case that passed on its suite's previous run and fails now. |
| Run verdict | Clean, Tuned, Drifted or Failed — how a test case came out of a tuned run. |
| Script-defect advisory | A warning that a step passed despite wording that no longer matches the application. |
| Script fingerprint | A SHA-256 code identifying a test case exactly as written when the run started. |
| Self-healing | An automatic single retry with adapted tactics after a failure. |
| Suite | A saved collection of test cases run together. |
| Token | {{data:Sheet.Column}} — a placeholder replaced by a value from the bound dataset row. |
| Verdict | PASS, FAIL or ERROR. |
| Verification coverage | The share of a run's steps whose expected result was checked and met. |
| Verified rewrite | A suggested wording confirmed to identify exactly one control on the live page. |
