skill-transfer

Safely migrate AI coding agent configurations between different tools. Switch between Claude Code, OpenCode, and Kilo Code without losing your settings, permissions, or MCP server configs. Scan, diff, plan, and apply migrations with built-in rollback support.

Cognitive ARC-AGI-3 solver: 6 human-like drives, 10 reasoning modes (incl. simulation physique), 20 domaines physiques, micro-NN experts (580 params), multi-agent architecture.

Zero-hidden neural networks that solve non-linear problems through temporal depth, not spatial layers. 90.14% MNIST with 480 parameters. Intelligence is not depth — it's resonance. Time is the ultimate hidden layer. OdyssNet proves it.

This project investigates how visual feedback influences throwing accuracy in virtual reality (VR), aiming to understand and enhance the transfer of VR training to real-world skills.

SILT is a hardware-aware skill-transfer and trust layer for AI. It is designed to let one AI acquire a specialist capability from another AI, prove that the capability works on unseen tests, and adapt the resulting system for constrained hardware without silently sacrificing the skills that matter.