AutoReg
AutoReg is an Electron desktop app that reads your autoreg.md spec, opens a real browser, and generates a complete Playwright E2E test suite — then runs, analyzes, and auto-heals failing tests with a LangGraph AI agent.
What AutoReg does
| Step | What happens |
|---|---|
| 1. Write spec | Describe your app’s features and test scenarios in autoreg.md |
| 2. Generate | The AI agent opens your app in a headless browser, captures live DOM snapshots, and generates scenario-NNN.json test files |
| 3. Plan | Tests are grouped into logical sections in plans/e2e-plan.json |
| 4. Run | Execute individual tests or the full suite with one click |
| 5. Autoheal | Failing tests are analyzed, root-caused, and regenerated automatically |
Core features
- Browser-grounded generation — uses real ARIA selectors from live DOM, not guessed selectors
- Live progress — accordion panel updates as each test is generated
- JSON Step DSL — portable interchange format; readable and editable
- Autoheal loop —
test_run_analyzer→test_regenerator→apply_scenario_fix - Encrypted API key storage — OS keychain on Mac, DPAPI on Windows
- Model switching — change OpenAI model in the chat composer without restart
Quick navigation
- Installation — prerequisites, install, API key setup
- Writing autoreg.md — spec format, sections, tips
- Generating tests — create project and run the 3-step agent
- Running and managing tests — execute, review, edit scenarios
- Autohealing — analyze failures and regenerate broken tests