[devin-ai-integration]

Summary

Automates qa/manual-test-plan.csv as Playwright tests (e2e/) that run in CI against the real app on a MySQL 8 service container (.github/workflows/e2e.yml). 54 of 60 rows automated; the 6 that aren't are listed with reasons in qa/coverage-matrix.md, which covers all 60 rows.

• Each test title starts with its plan ID (test('MT-031 Overdraw is rejected', ...)). e2e/scripts/check-coverage.mjs runs in CI and fails if a plan row is missing from the matrix, an Automated row has no matching test (or more than one), or a Not automated row has no reason.
• Each test registers its own unique customers, so there are no shared fixtures and tests can run in parallel. Local run: 54 passed in 17.7s against MySQL 8 (demo/start-demo.sh).
• "Form is not submitted" rows (MT-025/026/034/043/044) record outgoing POST requests and assert there were none, plus check the HTML5 validity flag. The balance check alone isn't enough to prove it.
• Dashboard panels are Bootstrap collapses loaded from a CDN, so openPanel waits for jQuery.fn.collapse before clicking. Without that wait the tests race in CI.
• CI builds with ./mvnw -s demo/maven-settings.xml (GCS mirror) and uses mirror.gcr.io/library/mysql:8.0, as AGENTS.md requires. It also asserts that the server version is 8.*.

Not automated (6):

ID	Why
MT-021	The expected result is a brand-team sign-off against a brand guide that isn't in the repo, so there's no pass condition a machine can check.
MT-047	The expected result is "System handles it appropriately", which isn't a defined behaviour, so I didn't invent one. What happens today: a -50 transfer is accepted and takes $50 from the recipient (sender $100 → $150, recipient $0 → $-50). A -50 deposit is also accepted. This needs a product decision.
MT-048	The app has no SMS or email capability, so there's nothing to observe.
MT-055	Needs a physical branch printer and a person to judge the printout.
MT-059	Needs JAWS on Windows and a person to judge whether announcements are clear. An axe scan wouldn't be equivalent.
MT-060	Moderated usability session with a real first-time customer.

Test precondition I added: MT-042 deposits 100 first. transferAmount checks the balance before it looks up the recipient, so a $0 sender would get "Insufficient funds" instead of "Recipient account not found".

Link to Devin session: https://devin-gtm.devinenterprise.com/sessions/5408d659a20a42ebb3c4cd2c8b086de9
Open in Devin Desktop: https://devin-gtm.devinenterprise.com/desktop/session/5408d659a20a42ebb3c4cd2c8b086de9?variant=devin
Requested by: @GarretAnd

---

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.

[devin-ai-integration]

Closed by demo/seed-situation.sh: demo_qa seed branch was reset.