The latest articles on DEV Community by Broxhq (@broxhq). https://dev.to/broxhq https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3966385%2F22f7c468-e38e-4580-9511-7f7137d938df.png https://dev.to/broxhq en Broxhq Wed, 03 Jun 2026 11:00:50 +0000 https://dev.to/broxhq/i-built-an-ai-agent-that-runs-manual-test-cases-in-a-real-browser-4caj https://dev.to/broxhq/i-built-an-ai-agent-that-runs-manual-test-cases-in-a-real-browser-4caj <h2> The problem </h2> <p>Every deploy — same manual test steps. Login, open the form, <br> fill the fields, check the result. Over and over.</p> <p>I wanted to skip the Playwright/Selenium boilerplate and just <br> paste my existing test cases as plain text.</p> <h2> What I built </h2> <p>qpilot — an AI agent that reads your manual test case and <br> executes it in a real Chrome browser step by step.</p> <p>You write this:</p> <ol> <li>Go to <a href="https://myapp.com/login" rel="noopener noreferrer">https://myapp.com/login</a> </li> <li>Enter email and password</li> <li>Click Login</li> <li>Verify dashboard is visible</li> </ol> <p>The agent opens Chrome, clicks, fills forms, and reports <br> pass/fail/warn per step with evidence from the page.</p> <p>If it hits an OTP or captcha — it pauses and asks you directly.</p> <h2> How it works </h2> <ul> <li>Playwright controls the real Chrome browser</li> <li>Each step: snapshot → action → snapshot → report</li> <li>Claude Haiku reads the snapshot (ARIA tree) and decides what to click</li> <li>Element refs (e.g. e12) are used for precise targeting</li> <li>Context window is managed to avoid hitting token limits</li> </ul> <h2> Try it </h2> <p>npx qpilot</p> <p>No code. No config. No Selenium.</p> <h2> Stack </h2> <p>TypeScript, Playwright, Claude Haiku via Anthropic API.</p> <p>Open source: <a href="https://github.com/broxhq/qpilot" rel="noopener noreferrer">qpilot</a></p> <p>Curious what you think — especially about edge cases <br> you'd want it to handle.</p> ai automation showdev testing