Manual-first QA workflow for the ParaBank demo app, run through Claude Code: story analysis → AC validation → test plan → test cases → (only on request) Playwright automation.
For each user story:
- Analyze the story and flag ambiguities
- Validate AC — completeness, testability, missing edge cases
- Test plan / strategy
- Test cases — ID, preconditions, steps, expected result, risk-based priority (High/Med/Low)
- Automation — Playwright, only run explicitly or via
/run-tests
This is enforced by CLAUDE.md and the Claude Code skills in .claude/skills/.
| Path | Purpose |
|---|---|
stories/ |
Source user stories (input) |
features/ |
Gherkin .feature files derived from stories |
test-artifacts/ac-validation/ |
AC validation notes per story |
test-artifacts/test-plans/ |
Test strategy/plan per story |
test-artifacts/test-cases/ |
Test case tables per story |
test-artifacts/execution-logs/ |
Output from automated test runs |
test-data/ |
Test data fixtures (e.g. users.json) |
.claude/skills/ |
manual-test-workflow, execute-registration, execute-browser-flow, code-review |
.claude/commands/ |
Custom slash commands (/run-tests) |
index.html |
Legacy Tic-Tac-Toe demo, unrelated to the QA workflow — kept from before the repo's pivot |
customer-registrationcustomer-loginfund-transfer
Each has AC validation, a test plan, and test cases under test-artifacts/.
- Registration: unique username + SSN generated per run (timestamp-based)
- Login: fixed seed account for standalone login tests
- Register → login chain: credentials passed in-run from registration to login, never hardcoded
- Drop a new story into
stories/. - Ask Claude Code to run the manual test workflow — it produces AC validation, test plan, and test cases in
test-artifacts/. - Once test cases are finalized, run
/run-tests(or ask explicitly) to automate them with Playwright.