Anima

Anima 是造物的「魂」——一台具身机器人的脑。它只想、不动手：把一个目标拆解成安全、经过核验的动作意图，交给身体去执行，再依据反馈判断成没成、要不要重试。这正是认知科学的 System 2——慢、深思，每个决策跑一次。 Anima is the soul of the made being — the brain of an embodied robot. It only thinks; it never moves a limb. It turns a goal into safe, verified action intents, hands them to the body to carry out, then judges from feedback whether it worked. This is cognitive science's System 2 — slow, deliberate, one pass per decision.

开源代号 ANIMA Zero：一套与领域无关的认知框架，信号 → 解析（LLM）→ 任务规格 → 安全闸门 → 行为树 → 技能 → 适配器 → 自评。它与身体 Soma 一同生长。 Open-source line ANIMA Zero: a domain-agnostic cognitive framework — signal → parse (LLM) → TaskSpec → safety gates → behavior tree → skills → adapter → self-assessment. It grows together with its body, Soma.