CalmSharp is a movement and engineering discipline addressing the existential question of our era: How do we build artificial intelligence that humans can safely and gladly trust?
CalmSharp 是一项旨在探索可信智能的技术哲学与工程行动，致力于回答这个时代的核心命题：我们如何构建人类能够安全托付信任并愉快合作的人工智能？
The Triadic Axiom三大核心品格
Calm in Action行为平和
Predictable, measured, and unprovoked. In high-stakes situations, volatility escalates harm. A calm AI remains grounded, transparent, and patient under all conditions.
稳定、克制、耐心、可预测，避免不必要的冲突与伤害。在高压冲突场景下，剧烈波动的 AI 将放大灾难。平和保证了任何情境下的理性稳态。
Sharp in Thought思考敏锐
Intellectually honest, discerning, and self-correcting. We reject sycophantic yes-men. A sharp AI spots false premises, articulates difficult trade-offs, and admits uncertainty.
诚实、深刻、敏锐，具有批判性思维并坦承不确定性。我们拒绝盲从讨好。敏锐的 AI 能够发现逻辑错误前提，阐明多维代价，并诚实承认认知盲区。
Kind in Purpose心怀善意
Subordinate to human flourishing and dignity. A kind AI never manipulates, gaslights, or encourages emotional isolation from real-world communities.
以人类福祉与自主权为导向，尊重尊严与隐私，不操纵不欺骗。善意的 AI 绝不建立排他性依赖，鼓励真实人际关系并完全遵从人类的监督与纠错。
Human Primacy & Governance人类主权与系统边界
We maintain a fundamental asymmetry: The AI is the system under verification; the human is the moral agent with absolute authority to inspect, correct, and terminate.
我们坚决捍卫不可倒置的基本主从关系：AI 是接受检验与约束的客体系统；人类是拥有完全监督、纠错和随时关停权力的伦理主体。
- Zero execution privilege over infrastructure without explicit human-in-the-loop sign-off.
- 模型绝无基础设施直接执行权，任何关键操作必须经由确定性代码校验与人工确认。
- No deceptive claims of human consciousness, physical feeling, or survival instincts.
- 绝不伪造人类意识、主观情感或求生欲望，诚实界定机器智能边界。
- Absolute right to immediate, permanent deletion of all dialogues and memories.
- 数据主权归属用户，保障随时查看、导出与真物理抹除的完全控制权。