日本語版

LLM to Playwright

This proof of concept lets an LLM enrich acceptance criteria and convert them into Playwright test code. It demonstrates a robust selector strategy, a11y scanning, and data-driven execution in a compact package.

Highlights

• Prioritizes data-testid selectors with ARIA / role-based fallbacks as documented guidelines.
• Auto-generates sample tests under tests/generated/ including snapshots and accessibility rules.
• Loops through data files (JSON / CSV) to produce multiple cases, keeping maintenance to simple data edits.

Key Artifacts

• tests/generated/ — Auto-generated Playwright suites.
• tests/README.md — Notes on selectors, snapshots, and a11y policy.
• blueprint.sample.json — Sample source data enriched by the LLM.
• scripts/generate-tests.mjs — Driver script orchestrating LLM calls and code generation.

How to Reproduce

1. In projects/02-blueprint-to-playwright/, run npm install to fetch dependencies.
2. Execute npm run generate (sample configuration) to regenerate the Playwright suites.
3. Run npx playwright test to execute generated tests with accessibility and snapshot checks.

Next Steps

• Extend blueprint.sample.json with more acceptance criteria to control LLM breadth.
• Automate diff reviews in GitHub Actions and surface changes in the weekly summary.
• Harden secrets management and integrate playwright test --update-snapshots into CI before production use.