What problem does it solve? Iterating on a web game without automated feedback leads to broken controls, visual regressions, and untested state transitions. This Skill enforces a tight implement-act-pause-observe loop so every change to an HTML/JS game is validated with real browser input, screenshots, and console error checks. ## Core Features & Use Cases - Playwright action loop: Run scripted keyboard and mouse bursts against the game via a dedicated client script, with deterministic frame stepping through a window.advanceTime(ms) hook. - State and visual inspection: Capture canvas screenshots and window.render_game_to_text JSON output after each burst, plus buffered console error logs, so you can verify gameplay without manual play. - Structured iteration workflow: Track work in progress.md, follow a test checklist covering movement, win/lose transitions, collisions, and menus, and fix the first new console error before continuing. - Use Case: You are building a canvas platformer and just added a shooting mechanic. Run the client with an action payload that moves, jumps, and fires, then inspect the screenshots and state JSON to confirm enemies lose health, disappear at zero, and update the score. ## Quick Start Ask the agent to build or modify the web game in this workspace and validate each change by running the Playwright client script with an action payload, then reviewing the screenshots and render_game_to_text output.
