What problem does it solve? Playing Pokemon through an emulator requires constant visual attention and manual input, making it impossible to automate or demonstrate AI gameplay without a structured control loop. This Skill provides a complete workflow for running Pokemon games headlessly, reading game state from emulator RAM, and making vision-verified decisions. ## Core Features & Use Cases - Headless Emulation Server: Runs the pokemon-agent package with PyBoy to serve game state, screenshots, and action endpoints over HTTP on port 9876. - Vision-Guided Gameplay Loop: Combines RAM state reads (position, HP, battle, dialog) with screenshot analysis to navigate maps, handle battles, and avoid getting stuck on ledges or warps. - Save State Management: Creates named save states before risky encounters and resumes from them via API or startup flags. - Live Dashboard Sharing: Sets up an SSH reverse tunnel through localhost.run so users can watch gameplay in a browser. - Use Case: A user says "play pokemon" and the AI starts Pokemon Red, navigates from Pallet Town to defeat Brock, saving progress before the gym battle while the user watches via a shared dashboard URL. ## Quick Start Ask the AI to "play pokemon" and provide the path to your own legally obtained Game Boy ROM file.
