An AI-powered quality engineering system built on Claude Code. It covers the full quality lifecycle — from raw requirement documents all the way to runnable, production-ready test automation artifacts — across three self-contained capability modules.
Capability 1 — req-to-story/ Requirement Doc → User Stories
Capability 2 — story-to-testcase/ User Story → Manual Test Cases
Capability 3 — tc-to-automate/ Test Cases → Automation Artifacts
Start at any point in the pipeline — or chain all three together:
Req Doc
→ /gen-user-stories
→ User Stories
→ /gen-manual-tests
→ Manual TCs (BDD .feature or non-BDD .md)
→ /generate-automation-tests
→ POMs + Spec Files + BDD Artifacts + CI Pipeline
Or run the full chain in one command:
/gen-user-stories --req inputs/req-docs/prd.pdf --then-e2e --url https://staging.myapp.com
| Artifact | Location |
|---|---|
User story files (US-NNN.md) |
outputs/<project>/user-stories/ |
Manual TC documents (BDD .feature or non-BDD .md) |
outputs/<project>/manual-tests/manual-tcs/ |
| Page Object Model classes | outputs/<project>/src/pages/ |
Playwright spec files (*.spec.ts) |
outputs/<project>/tests/nonbdd/ |
| Gherkin feature files + step definitions | outputs/<project>/tests/bdd/ |
| Test data fixtures | outputs/<project>/src/fixtures/ |
playwright.config.ts + package.json + tsconfig.json |
outputs/<project>/ |
| GitHub Actions CI pipeline | outputs/<project>/configs/.github/workflows/ui.yml |
| Allure + Playwright HTML reports | outputs/<project>/allure-report/ |
| Tool | Version | Install |
|---|---|---|
| Node.js | 18.x LTS+ | nodejs.org |
| Git | 2.x+ | git-scm.com |
| Claude Code CLI | Latest | npm install -g @anthropic/claude-code |
| Java | 8+ | Required for Allure CLI only |
git clone <repo-url> sqe-kit
cd sqe-kit
claude # or open in VS Code with the Claude Code extension/setup-kit
Select kit-u1 (Playwright TypeScript). This installs dependencies and creates the output
directory structure.
# From a live URL (best locator accuracy)
/generate-automation-tests --url https://practice.expandtesting.com/login --cases "TC-01: Successful Login
Steps:
1. Navigate to the login page
2. Enter username: practice
3. Enter password: SuperSecretPassword!
4. Click Login
Assertions:
- URL changes to /secure
- Flash message contains 'You logged into a secure area!'"
# From a requirement document
/gen-user-stories --req inputs/req-docs/your-spec.pdf
# From a user story
/gen-manual-tests --story inputs/user-stories/your-story.md
# Via kit commands (runs both Playwright specs + Cucumber BDD)
/run-smoke # @smoke tagged tests only
/run-test # all tests
# Directly
cd outputs/playwright-greenfieldproject
npx playwright test # Playwright specs
npx cucumber-js # BDD feature files
npx allure open allure-report # View Allure reportsqe-kit/
├── kit.config.json ← Active kit (switch kits by changing "id")
├── CLAUDE.md ← Orchestrator — all routing + quality rules
│
├── req-to-story/ ← Capability 1 (kit-independent)
│ ├── agents/ ← req-ingestor, req-analyzer, feature-splitter, user-story-writer
│ ├── skills/ ← req-text-extractor
│ └── rules/ ← user-story-quality
│
├── story-to-testcase/ ← Capability 2 (kit-independent)
│ ├── agents/ ← quality-master-orchestrator, 8 specialist agents, gap-filler
│ ├── skills/ ← user-story-parser, tc-formatter-bdd, tc-formatter-nonbdd
│ └── rules/ ← manual-tc-quality
│
├── tc-to-automate/ ← Capability 3 (kit-aware)
│ ├── kits/
│ │ ├── _base/agents/ ← intake, locator, pom, testgen, bdd, datagen, ci, reporting
│ │ └── kit-u1/ ← Playwright TypeScript (KIT.md + templates)
│ ├── configs/integrations.yml ← Plugins: Jira, GitHub, Allure Server
│ └── rules/ ← playwright-ts, page-objects, bdd-gherkin, selenium-java, python-test
│
├── inputs/
│ ├── req-docs/ ← Place requirement documents here
│ ├── user-stories/ ← Place user story files here
│ ├── snapshots/ ← Place HTML DOM snapshots here (Tier 2 locators)
│ └── test-cases/ ← Place test case files here
│
├── plugins/ ← github.md, jira.md, allure-server.md
│
├── docs/
│ ├── TECHNICAL-REFERENCE.md ← Full technical reference for kit developers
│ └── USER-GUIDE.md ← Step-by-step user guide
│
└── outputs/
└── playwright-greenfieldproject/ ← All generated artifacts
| Command | Capability | What it does |
|---|---|---|
/gen-user-stories |
1 | Requirement doc → US-NNN.md user story files |
/gen-manual-tests |
2 | User story → manual TCs (BDD .feature or non-BDD .md) |
/e2e |
2 + 3 | User story → manual TCs → automation artifacts |
/generate-automation-tests |
3 | Test cases → full automation artifact set |
/add-test |
3 | Add a single test to an existing project |
/gen-bdd |
3 | Generate BDD feature + step definition files only |
/generate-locators |
3 | Extract locators → selectors.json only |
/generate-page-objects |
3 | Build POM files from existing selectors.json |
/ingest |
3 | Normalize inputs → intake.summary.json only (dry run) |
/run-smoke |
— | Run @smoke tests (Playwright + Cucumber) + generate reports |
/run-test |
— | Run all tests (Playwright + Cucumber) + generate reports |
/setup-kit |
— | Scaffold project structure + install dependencies |
/setup-reporting |
— | Configure Allure + HTML reporting |
/setup-ci |
— | Generate GitHub Actions CI pipeline |
/validate-kit |
— | TypeScript compile + test discovery check |
/gen-data |
— | Generate test data fixtures |
The kit selects locators using three tiers in priority order:
| Tier | Strategy | Trigger | Score cap |
|---|---|---|---|
| 1 | Playwright MCP (live browser) | --url provided |
None — up to 1.0 |
| 2 | DOM snapshot parsing | --dom provided |
None — up to 1.0 |
| 3 | AI-guided inference | Neither --url nor --dom |
0.75 — always verify |
Always use Tier 1 or 2 when possible. Tier 3 locators are flagged in selectors-lint.html
and must be verified against the real application before running in CI.
Kit selection is controlled by kit.config.json. Run /setup-kit after changing it.
| Kit ID | Stack |
|---|---|
kit-u1 |
Playwright + TypeScript (active) |
kit-u2 |
Playwright + TypeScript + Cucumber (BDD-first) |
kit-u3 |
Selenium + Java + TestNG |
kit-u4 |
Selenium + Java + Cucumber |
kit-a2 |
pytest + requests (API) |
kit-h1 |
Playwright + TypeScript + Hybrid (API + UI) |
kit-h3 |
Selenium + Python + Hybrid |
.pdf, .md, .txt, .docx, .xlsx, .yaml, .json, .xml, .bpmn, .csv, .eml, .png, .jpg
.md, .txt, .pdf, .yml — or inline text in any of these formats:
- Agile:
As a <role>, I want <goal>, so that <benefit> - Plain English:
# Feature: X+ description paragraphs - Structured:
## Feature+**Acceptance Criteria**sections
Plain-English numbered steps, Gherkin-like syntax, OpenAPI spec (JSON), Postman collection (JSON), or .md files
- Locators are never inline in spec files — always in Page Object classes
- BDD artifacts are always paired (
.feature+.steps.tsin same run) waitForTimeout()is forbidden — useexpect(locator).toBeVisible()- Every BDD scenario must have
@TC-NNN,@ui/@api, and@smoke/@regressiontags - Selector file is always
selectors.json— neverlocators.json - Generation is idempotent — existing files are never overwritten without confirmation
| Document | Contents |
|---|---|
| docs/USER-GUIDE.md | Step-by-step guide for first-time users |
| docs/TECHNICAL-REFERENCE.md | Architecture, agents, skills, rules, config reference |
Kit version: 2.0 | Active kit: kit-u1 (Playwright TypeScript) | Last updated: 2026-03-13