a fast, tiny, extensiable workflow engine
Integer types that cannot be the maximum value, allowing for memory layout optimization and intuitive arithmetic operations.
System 1 relay for LLM apps — sub-millisecond intent classification, safety gating, tool selection. CPU-only, continuous learning from corrections.
Iteratively stack mutable references
Rust implementation of Wave Function Collapse (WFC) algorithm with type safety and direction-aware grid system
Pixel art game asset generator. Stardew/Starbound quality. Local-first, no cloud.
Reusable AI assistant library for local LLM integration (Ollama, LM Studio, etc.)
A small example before developing computer chess and computer shogi. Come see the repository.
WIP. Connect-four AI. Currently, does not work.
Kova — augment engine. On-device AI inference, agentic tool loop, distributed C2 mesh. Zero cloud. One binary.
MCP server for VDSL — AI-native image generation platform. RunPod GPU provisioning, ComfyUI orchestration, model management.
File synchronization engine — N-location, pluggable store & backend