Test Generation
Test Baseline Review
Learn how to review, approve, and reject AI-generated test cases to build a verified test baseline for your regression suite.
Test Baseline Review
AegisRunner has two baseline concepts that work together:
- Test Suite Baseline — Review and approve AI-generated test cases before they run in CI
- Crawl Baseline — Set a completed crawl as the reference for regression testing
Test Suite Baseline
When AI generates test cases, each starts with a pending review status. Review them to build a trusted test baseline.
Review Status
| Status | Meaning |
|---|---|
| Pending | Not yet reviewed — will still run but flagged as unreviewed |
| Approved | Verified as correct — part of your trusted baseline |
| Rejected | Marked as incorrect — disabled from future runs |
Review Workflow
- Open a test suite and view its test cases
- For each test: review the steps, selectors, and assertions
- Click Approve if the test is correct, or Reject if it is wrong
- Rejected tests are automatically disabled and won't run in CI
Stale detection: When you re-crawl a site and tests are regenerated, existing approved tests are marked as needs review so you can verify the updated tests still match your expectations.
Crawl Baseline (Regression Manifests)
A crawl baseline captures a known-good state of your site for deterministic replay.
Setting a Crawl Baseline
- Complete a Full Site crawl
- On the crawl results page, click Set as Baseline
- AegisRunner compiles a regression manifest — the exact pages, interactions, and expected states
- The Regression crawl mode becomes available
What Gets Compiled
- All discovered pages in crawl order
- All interactions per page (click actions, form submits, dropdowns)
- Expected state count per page
- DOM hashes for state comparison
Running a Regression Crawl
From the crawl page, select Regression mode. The crawler replays the manifest step-by-step with a single worker for maximum determinism. Any differences from the baseline are flagged as regressions.
Comparing with Baseline
After a crawl completes, click Compare with Baseline to see:
| Category | Meaning |
|---|---|
| Missing | Pages in baseline not found in current crawl — potential regressions |
| Changed | Pages with different state counts or interactions |
| New | Pages found in current crawl but not in baseline |
| Unchanged | Pages matching baseline exactly |
See Regression Manifests for the full manifest structure and policy options.
Related Documentation
- Regression Manifests — Deterministic replay testing
- Visual Regression Testing — Pixel-level screenshot comparison
- AI Test Generation — How tests are generated