Flesh out E2E Playwright tests and run them in CI

[rustydb]

Priority: P2

Source

Senior Architect Code Review (2026-04-03)

Problem

Four Playwright E2E spec files exist under apps/dashboard/tests/e2e/ but they are minimal stubs (284-420 bytes each). More critically, E2E tests are not run in CI — the CI workflow only runs bun run test:dashboard (Vitest unit tests), not bun run test:e2e (Playwright).

The Playwright config is set up but the specs need to be fleshed out to provide actual browser-level regression coverage.

Affected Files

• apps/dashboard/tests/e2e/dashboard.spec.ts
• apps/dashboard/tests/e2e/detail.spec.ts
• apps/dashboard/tests/e2e/map.spec.ts
• apps/dashboard/tests/e2e/performance.spec.ts
• apps/dashboard/playwright.config.ts
• .github/workflows/ci.yml

Recommendation

1. Flesh out the E2E specs to cover critical user journeys:

   • Wallet connect / disconnect flow
   • Turret card selection and detail panel interaction
   • Network node drawer open/close
   • Filter application and clearing
   • Demo mode navigation and rendering

2. Add a Playwright CI job that:

   • Starts the dev server (or builds and serves)
   • Runs E2E tests against the running app
   • Uploads trace artifacts on failure

Acceptance Criteria

• E2E specs cover at least 5 critical user journeys
• CI has a dedicated E2E job that runs Playwright tests
• Test traces are uploaded as artifacts on failure
• E2E tests pass reliably (no flaky timing assertions)
• Playwright browsers are cached in CI

[m13v]

we had the exact same situation, stub playwright specs sitting there for months while only vitest ran in CI. the thing that actually made us flesh them out was starting with just one critical flow instead of trying to cover everything at once. wallet connect would be a good first one since it touches the most infrastructure. one gotcha with playwright in CI specifically, make sure you're running with --reporter=html and uploading the artifacts, otherwise when tests fail on a PR you're blind debugging. also sharding the tests across browsers early saves a lot of pain later vs running them sequentially.

[m13v]

fwiw we open sourced our test generation approach at assrt.ai. it crawls your running app, discovers the user flows automatically, and generates real Playwright test files. the CI integration and HTML reporting is handled out of the box. the core test discovery logic: https://github.com/assrt-ai/assrt-sdk