Local inference server optimized for consumer hardware that profiles available resources, recommends suitable models, and connects to existing coding agents such as OpenCode and Claude Code.