Test Generator

Describe what to test in plain English or paste a Gherkin scenario. Jerry generates executable Playwright TypeScript — then runs it.

▶How this works

Mechanism

Your requirement is sent to Claude (Haiku) with a system prompt that instructs it to output only valid Playwright TypeScript. The response streams back token-by-token via Server-Sent Events — you see the code appear in real time as the model generates it.

// System prompt excerpt (Natural Language mode)
You are an expert QA engineer. Given a plain-English requirement,
output ONLY valid Playwright TypeScript test code using @playwright/test.
Use data-testid selectors where possible. No explanation, only code.

Gherkin / BDD mode

In Gherkin mode the system prompt changes: Claude is asked to parse the Feature/Scenario/Given/When/Then structure and map each step to a Playwright action. This bridges BDD specs directly to executable tests without a step-definition layer.

Run Test

Clicking "Run Test" POSTs the generated code to /api/tests/run-generated. The server prepends // @ts-nocheck, writes it to a temp file in playwright/generated/, calls the same Playwright runner used by the Test Runner page, streams results back as SSE, then deletes the file.

Why this matters for QE: eliminates the gap between "I know what to test" and "I have a passing spec". A non-technical stakeholder can describe behaviour in plain English and get a runnable Playwright test in under 10 seconds.

input

Requirement

examples

generated_test.ts

Awaiting input

Generated Playwright TypeScript will appear here