AISquare Studio AutoQA
A GitHub Action that turns natural-language test steps written in a pull request description into generated, executed, and committed Playwright test code, using multi-agent orchestration.
AISquare Studio AutoQA converts plain-English test steps, written inside a fenced code block in a pull request's description, into fully automated Playwright Python tests. When a pull request is opened, edited, or synced, the GitHub Action parses the PR body for metadata such as flow name, criticality tier, and area, then uses CrewAI agents together with OpenAI GPT-4 to generate the test code step by step, using a DOM-inspection tool to auto-discover the right selectors from the live page and retrying automatically with alternative selectors on failure.
Generated code is validated with AST-based analysis before execution to block unsafe code patterns, then executed against the project's staging environment; on success, the test file is committed to a tiered folder structure, and results, screenshots, an HTML report, and a JSON artifact are posted back as a PR comment. An idempotency check avoids regenerating tests for an unchanged PR description, and pip and Playwright browser caching are used to keep CI runs fast.