| name | playwright-ai-mcp-tutor |
| description | Teach and implement Playwright AI agent workflows (Planner, Generator, Healer) with MCP and repository guardrails. Use when preparing workshops, generating Playwright tests with architecture constraints, or enforcing deterministic code generation through a rule engine. |
Playwright AI MCP Tutor Skill
Use this skill to teach or execute AI-assisted test automation in this repository.
Core Workflow
- Read architecture constraints from
references/rules.md.
- Ask Planner to output scenarios, tags, and impacted files only.
- Ask Generator to write code only in approved folders.
- Run
npm run rules:check and relevant tests.
- If failing, ask Healer to patch minimal code using logs/traces.
- Re-run rule engine and tests.
Required Constraints
- Keep
Pages -> Modules -> Tests layering.
- Place page objects in
src/pages.
- Place module/modal workflow code in
src/modules.
- Place utility code in
src/utils.
- Keep direct locators out of modules.
Use Prompt Templates
Load references/prompts.md and reuse prompts with repository-specific inputs.
Teaching Sequence
- Explain the three-agent model.
- Demonstrate MCP browser grounding.
- Run a live generate -> validate -> heal loop.
- Make students pass
rules:check before grading.