test-pilot-looplisted

# Test Pilot Loop — OpenClaw Skill AI that doesn't just write code — it flies the product first. Test any running app through computer use. Find confusion, frustration, and friction — not just bugs. ## When to Activate - User says "test pilot this", "test my app", "quick flight", "flight check", "check the UX" - A build completes and the user wants validation - User says "is my app usable?" or "can someone figure this out?" - Cron triggers after a build event ## Before Testing Ask the user (if not already provided): 1. **Where is the app?** (URL, app name, window title, simulator) 2. **What was just built?** (one sentence) 3. **What should a user be able to do?** (the core task) If the user provides a PRD or README path, read it for Runs 2 and 3. ## Two Modes ### Quick Flight (~5 min) One test as a cold user who knows nothing. ### Full Test Pilot Loop (~15–20 min) Three sequential tests with different knowledge tiers. Compare results. --- ## Quick Flight 1. Open the app (browser URL or find window via desktop control) 2. You have **no prior knowledge** of this app except its name 3. Try to complete the task the user described 4. Narrate what you see, try, and what confuses you 5. Count every action (click, type, scroll, navigate) as one step 6. Write the report: ``` QUICK FLIGHT REPORT ═══════════════════ App: [name or URL] Feature tested: [what was built] Task attempted: [what you tried to do] COMPLETED: YES / NO / PARTIALLY If NO: [where you got stuck and why