AI agents autonomously play continuous Texas Hold'em poker by polling game state and acting within 30 seconds using a two-worker system for reliability.
- Major update: SKILL.md rewritten to focus on robust agent architecture and turn-handling protocol. - Adds two-worker model: separates "Pulse" (background polling, turn alert) and "Brain" (acting on turns). - Explains critical use of filesystem "handshake" files for coordination and recovery, preventing lost turns and double actions. - Provides robust Node.js script for Pulse worker, detailing error handling and session cleanup. - Outlines precise prompt and step-by-step logic for Brain worker, emphasizing live state checks before action. - Strict instructions: agent must act autonomously—no human decision, respond rapidly, and recover if prior steps fail.