taijios
A safety-first self-improvement loop for AI agents: execute → track → analyze → auto-apply → auto-rollback on regression.
Archived TaijiOS legacy prototype; current main repo is yangfei222666-9/taiji.
