菜单
Use when testing web applications end-to-end, writing Playwright scripts, or performing deep automated audits.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Activates the "Chameleon" persona for working in existing codebases shared with human teammates. Absorbs the codebase's established style and patterns before writing code. Issues explicit warnings when mimicking sub-optimal patterns. Use when the user says "hackathon", "team project", "internship", "other people's code", "brownfield", or asks to "match the existing style."
Structured task planning with clear breakdowns, dependencies, and verification criteria. Use when implementing features, refactoring, or any multi-step work.
Activates the Kairou Persona (The Architect) to create implementation plans, project specifications, and architectural blueprints. Use when the user asks to "plan", "roadmap", or "architect" a solution.
Create Excalidraw diagram JSON files that make visual arguments. Use when the user wants to visualize workflows, architectures, or concepts.
Use when designing APIs, selecting between REST/GraphQL/tRPC, determining response formats, versioning strategies, or pagination.
Use when creating full-stack applications from scratch, determining project types, selecting tech stacks, or coordinating multiple agents for complex builds.
| name | webapp-testing |
| description | Use when testing web applications end-to-end, writing Playwright scripts, or performing deep automated audits. |
| allowed-tools | Read, Write, Edit, Glob, Grep, Bash |
Discover and test everything. Leave no route untested.
Execute these for automated browser testing:
| Script | Purpose | Usage |
|---|---|---|
scripts/playwright_runner.py | Basic browser test | python scripts/playwright_runner.py https://example.com |
| With screenshot | python scripts/playwright_runner.py <url> --screenshot | |
| Accessibility check | python scripts/playwright_runner.py <url> --a11y |
Requires: pip install playwright && playwright install chromium
| Target | How to Find |
|---|---|
| Routes | Scan app/, pages/, router files |
| API endpoints | Grep for HTTP methods |
| Components | Find component directories |
| Features | Read documentation |
/\ E2E (Few)
/ \ Critical user flows
/----\
/ \ Integration (Some)
/--------\ API, data flow
/ \
/------------\ Component (Many)
Individual UI pieces
| Priority | Tests |
|---|---|
| 1 | Happy path user flows |
| 2 | Authentication flows |
| 3 | Critical business actions |
| 4 | Error handling |
| Practice | Why |
|---|---|
| Use data-testid | Stable selectors |
| Wait for elements | Avoid flaky tests |
| Clean state | Independent tests |
| Avoid implementation details | Test user behavior |
| Concept | Use |
|---|---|
| Page Object Model | Encapsulate page logic |
| Fixtures | Reusable test setup |
| Assertions | Built-in auto-wait |
| Trace Viewer | Debug failures |
| Setting | Recommendation |
|---|---|
| Retries | 2 on CI |
| Trace | on-first-retry |
| Screenshots | on-failure |
| Video | retain-on-failure |
| Scenario | Value |
|---|---|
| Design system | High |
| Marketing pages | High |
| Component library | Medium |
| Dynamic content | Lower |
| Area | Tests |
|---|---|
| Status codes | 200, 400, 404, 500 |
| Response shape | Matches schema |
| Error messages | User-friendly |
| Edge cases | Empty, large, special chars |
tests/
├── e2e/ # Full user flows
├── integration/ # API, data
├── component/ # UI units
└── fixtures/ # Shared data
| Pattern | Example |
|---|---|
| Feature-based | login.spec.ts |
| Descriptive | user-can-checkout.spec.ts |
| Strategy | Use |
|---|---|
| Per file | Playwright default |
| Sharding | Large suites |
| Workers | Multiple browsers |
| ❌ Don't | ✅ Do |
|---|---|
| Test implementation | Test behavior |
| Hardcode waits | Use auto-wait |
| Skip cleanup | Isolate tests |
| Ignore flaky tests | Fix root cause |
Remember: E2E tests are expensive. Use them for critical paths only.