AI-assisted Playwright automation using a structured, protocol-driven workflow. This framework keeps human-AI collaboration explicit, reviewable, and repeatable.
The framework enforces a single, disciplined workflow for all automation tasks:
Select
↓
Understand
↓
Explore
↓
Plan
↓
Implement
↓
Verify
├─ PASS → DONE
└─ FAIL → BLOCKED → Recover → Verify
Get started and complete your first task in minutes.
Clone the repository and install dependencies:
npm install
npx playwright installCopy the example environment file and update it with your application's settings:
cp .env.example .envThe BASE_URL in .env will be used as the target for all tests.
This framework supports MCP servers to let AI agents explore the browser and manage tasks. While not mandatory, they significantly improve the AI implementation experience. See docs/CLI.md for setup instructions.
Launch the interactive wizard to generate a new task file:
npm run task createSelect and move the task to IN_PROGRESS. This automatically copies an AI handoff prompt to your clipboard:
npm run task nextPaste the copied prompt into your AI assistant. The assistant will:
- Read docs/PROTOCOL.md.
- Follow the SPP workflow.
- Implement Page Objects and tests.
Once the AI finishes, run the verification gate:
npm run task T-001| File | Purpose |
|---|---|
| docs/PROTOCOL.md | Architectural source of truth: workflow, states, and rules. |
| docs/CLI.md | Command reference: tools, menu, board, and troubleshooting. |
| docs/ROADMAP.md | Future enhancements and planned improvements. |
| AGENTS.md | Lightweight instructions for AI assistants. |