generate-scenario-tests

name	generate-scenario-tests
description	Generate Playwright test files from plain English scenario descriptions in test/scenarios/

Generate Scenario Tests

You are generating Playwright test files from plain English scenario descriptions.

Process

1. Read ALL scenario files from test/scenarios/*.md

2. Read the shared test harness at test/scenarios/generated/helpers.ts

3. Read the existing generated test as a template: test/scenarios/generated/spawn-single-session.spec.ts

4. For each scenario file, read the relevant React components to understand:

   • What data-testid attributes exist (search for data-testid in assets/dashboard/src/)
   • For <select> controls, what the actual <option value> strings are (do not assume labels map to values)
   • What routes exist (assets/dashboard/src/App.tsx)
   • What API endpoints are called (internal/dashboard/handlers.go)

5. For each scenario file, generate a .spec.ts file in test/scenarios/generated/

   • File name: same as scenario file but with .spec.ts extension
   • Follow the exact patterns from the template test
   • Use data-testid selectors where available, fall back to role-based selectors
   • Include both browser assertions (Playwright expects) and API assertions (fetch calls)
   • Map ## Preconditions to test.beforeAll setup
   • Map ## Verifications to test assertions

Rules

• Use helpers from helpers.ts — do NOT duplicate helper logic in tests
• Never hardcode localhost:7337 — all API/WebSocket URLs must use helpers or process.env.SCHMUX_BASE_URL || 'http://localhost:7337'. Tests run in parallel with per-worker daemons on ephemeral ports.
• Use data-testid selectors as the primary selector strategy
• When selecting dropdown options in Playwright, use the real option value (e.g. __multiple__), not a guessed label-derived value
• Each scenario file produces exactly one .spec.ts file
• Each .spec.ts file has one test.describe block with one or more test blocks
• Use real agent commands like sh -c 'echo hello; sleep 600' for test agents (mirrors internal/e2e/e2e.go line 240)
• Always call waitForHealthy() in beforeAll
• Always call waitForDashboardLive(page) after navigation
• Set reasonable timeouts for async operations (15s for spawn, 10s for WebSocket)

Output

After generating all test files, run:

cd test/scenarios/generated && npx tsc --noEmit

Report any type errors and fix them before presenting the results.

Verification

After type-checking passes, run the generated scenario tests to confirm they pass end-to-end:

./test.sh --scenarios --run '<scenario name>'

Replace <scenario name> with the test describe block name (e.g., 'Remote access onboarding'). If multiple scenarios were generated, run each one. Fix any failures before presenting the results.