Architecture

Player Chat → Log Tail → AI Service → Ollama (Mortdecai) → RCON → Server

Components

• AI Service: Python process tailing the Minecraft server log
• Ollama: Local LLM inference server
• RCON: Remote console protocol for executing commands
• Validator: Post-processing that fixes syntax and enforces safety

Hardware

GPU	Role
RTX 4000 (16GB)	Production inference
RTX 3090 Ti (24GB)	Training + self-play
RTX 2080 Ti (11GB)	Self-play secondary