Turn NUnit history into ranked, explained flake evidence. Find unstable tests, manage expiring quarantine, and gate CI.A test that passes after a rerun is still a release risk. Test Flake Doctor imports the NUnit result files your Test Runner already creates, builds a committed history, and ranks the tests that do not decide consistently.What it helps you doImport existing Edit Mode, Play Mode, and CI NUnit XML without rerunning the suite.Score and rank likely flaky tests from repeated pass/fail evidence.Explain patterns such as order dependence, build-agent bias, duration instability, and failure-message churn.Run a focused flake hunt when one test needs repeated execution.Quarantine with an owner, reason, expiry, and visible overdue state.Redact account, machine, domain, and working-directory evidence from reports.Export deterministic JSON, Markdown, and self-contained HTML evidence.Import, analyse, report, and return a CI exit code in one batch run.Use it when reruns hide unstable tests, quarantine has become permanent, or a team needs evidence to decide what to fix first.The tool reports statistical and contextual evidence; it does not prove the root cause of nondeterminism. Imported history quality matters, and short histories remain low-confidence instead of being overstated.Read the complete documentation.PRODUCT TYPEEditor-only NUnit result history, flake diagnosis, quarantine, reporting, and CI tool.REQUIREMENTSUnity 2022.3 or newer, including Unity 6.NUnit-compatible XML result files from Unity Test Runner or CI.Unity Test Framework is optional for import/analysis and required for flake hunt and the quarantine attribute workflow.KEY FEATURESImports existing NUnit XML; normal analysis does not rerun tests.Committed, deterministic test history with stable test identity.Ranked flake scores with evidence strength and confidence language.18 diagnoses for sequence, agent, timing, message, and outcome patterns.Focused flake hunt for repeated execution of a selected test.Quarantine owner, reason, expiry, and overdue enforcement.Privacy redaction for account, machine, domain, and working paths.JSON, Markdown, self-contained HTML, and batch-mode CI output.IMPORTANT LIMITSEvidence indicates likely flakiness; it does not prove a code-level root cause.Sparse or low-quality history produces lower confidence.Does not replace Unity Test Runner or repair tests automatically.Flake hunt results can still be influenced by machine load and environment.Full documentation.AI-assisted image generation was used for the package icon artwork, and AI-assisted drafting was used for marketing copy and documentation formatting. All technical claims were sourced from the package, and the code, functionality, layouts, exports, and final text were manually reviewed and verified.
