By name: ui-audit description: Run automated UI/UX audit pipeline — 52 pages, 11 custom rules, axe-core accessibility, Storybook validation
UI/UX Audit — Automated Pipeline
Run the NOIR UI audit system: 52 pages, 11 custom rules, axe-core accessibility scanning, and Storybook story validation.
Step 1: Verify Prerequisites
Check that both backend and frontend are running:
curl -sf http://localhost:4000/api/health > /dev/null && echo "Backend: OK" || echo "Backend: NOT RUNNING"
curl -sf http://localhost:3000 > /dev/null && echo "Frontend: OK" || echo "Frontend: NOT RUNNING"
If either is not running, start them with ./start-dev.sh or manually:
- Backend:
cd src/NOIR.Web && dotnet run - Frontend (Windows):
powershell -Command "Start-Process cmd -ArgumentList '/c cd /d src\NOIR.Web\frontend && pnpm run dev'"
Optional — also check Storybook for story-level audit:
curl -sf http://localhost:6006 > /dev/null && echo "Storybook: OK" || echo "Storybook: NOT RUNNING (optional)"
Step 2: Run the Audit
cd src/NOIR.Web/frontend/e2e && npx playwright test --project=ui-audit --project=ui-audit-platform
This runs:
- ui-audit: 52 admin pages (tenant admin auth, includes
tenant-settings) - ui-audit-platform: 4 platform admin pages (platform admin auth)
- 11 custom rules: cursor-pointer, aria-label, datatable-actions, dialog-footer, badge-variant, empty-state, native-title, gradient-text, destructive-button, console-errors, network-errors
- axe-core: WCAG 2.2 accessibility scanning per page + per tab + per dialog
Output: .ui-audit/ directory (gitignored)
Data-load reliability (important)
Before taking screenshots or running rules, the audit runner must wait until pages are visually stable:
- Wait for navigation +
networkidle - Wait for app/API idle (no in-flight fetch/xhr calls for a short quiet window)
- Wait for page-specific selector (
waitForfrompage-registry.ts) - Wait for loading indicators/skeletons/spinners to disappear
If screenshots show empty table/list while data should be present, treat it as an audit-runner bug and improve readiness waits in tests/ui-audit/environment-setup.ts and runner specs.
Auth profile correctness (important)
Use the correct login profile per page:
ui-auditproject uses tenant admin (admin@noir.local)ui-audit-platformproject uses platform admin (platform@noir.local)
tenant-settings (/portal/admin/tenant-settings) must run in tenant-admin audit, not platform-admin audit. Keep page ownership/auth mapping in tests/ui-audit/page-registry.ts accurate to avoid wrong screenshots/tabs.
Step 3: Read and Present Results
After the audit completes, read the generated reports:
- Summary:
.ui-audit/summary.md— severity counts, top rules, worst pages - AI prompt:
.ui-audit/prompt.md— structured fix instructions grouped by source file - Raw data:
.ui-audit/raw/— JSON files for programmatic analysis
Present to the user:
- Total issue count by severity (CRITICAL / HIGH / MEDIUM / LOW / INFO)
- Top 5 rules with most violations
- Top 5 pages with most issues
- Ask: "Want me to fix CRITICAL and HIGH issues now?"
Step 4: Fix Issues (if requested)
Read .ui-audit/prompt.md and follow its instructions:
- Fix all CRITICAL and HIGH issues
- MEDIUM/LOW are recommended but optional
- Do NOT change UI behavior — only fix classes, components, attributes
- After all fixes:
cd src/NOIR.Web/frontend && pnpm run build
Step 5: Re-run to Verify (if fixes were applied)
cd src/NOIR.Web/frontend/e2e && npx playwright test --project=ui-audit --project=ui-audit-platform
Compare before/after issue counts. Goal: 0 CRITICAL, 0 HIGH.
Variant: Storybook-Only Audit
To audit only Storybook stories (axe-core on all 98+ stories):
cd src/NOIR.Web/frontend/e2e && npx playwright test --project=ui-audit tests/ui-audit/storybook-audit.spec.ts
Requires Storybook running at http://localhost:6006.
Variant: Single Page Audit
To audit a specific page by ID (from page-registry.ts):
cd src/NOIR.Web/frontend/e2e && npx playwright test --project=ui-audit -g "page: <page-id>"
Example: npx playwright test --project=ui-audit -g "page: products"