Patrick Morgan 分享了如何为 Sublime Security 构建专用原型环境，摆脱 Figma，实现团队协作与灵活性。
Patrick Morgan shares how he built a dedicated prototyping environment for Sublime Security, moving away from Figma to enable team collaboration and flexibility.
要点 · TL;DR
在独立于生产环境之外的原型制作环境中，迭代更快、更灵活。 Prototyping in a separate environment from production enables faster iteration and flexibility.
将原型集中存放在共享仓库中，团队可以逐步在彼此的工作基础上进行构建。 Centralizing prototypes in a shared repository allows teams to build on each other's work over time.
设计师应准备让 AI 代理重新诠释的设计，而非传统交接。 Designers should prepare designs for AI agents' reinterpretation, not traditional handoff.
核心观点 · Key points
在与生产环境分离的独立环境中进行原型设计，可以带来更大的灵活性和更快的迭代速度。 Prototyping in a separate environment from production allows for more flexibility and faster iteration.
将原型集中存储在共享代码库中，使团队能够随着时间的推移在彼此的工作基础上进行构建。 Centralizing prototypes in a shared repository enables teams to build on each other's work over time.
低保真原型对于早期探索很有价值，可以逐步提升到高保真度。 Low-fidelity prototypes are valuable for early exploration and can be gradually brought to high fidelity.
文档和设计原则应集成到原型设计环境中，以指导AI智能体。 Documentation and design principles should be integrated into the prototyping environment to guide AI agents.
设计师应准备设计以供工程师的智能体重新解读，而非传统的交接方式。 Designers should prepare designs for reinterpretation by engineers' agents rather than traditional handoff.
反共识 · Contrarian takes
设计师不应直接在生产代码中进行原型设计；独立环境更好。 Designers should not prototype directly in production code; a separate environment is better.
当AI智能体可以轻松地将更改移植回生产环境时，复制组件是可以接受的。 Duplicating components is acceptable when AI agents can easily port changes back to production.
在早期探索阶段，低保真原型可能比高保真原型更有效。 Low-fidelity prototypes can be more effective than high-fidelity ones for early-stage exploration.
与设计工具的主要交互应通过自然语言进行，而非直接操作。 The primary interaction with design tools should be through natural language, not direct manipulation.
内部工具可以由设计师在AI辅助下构建，而不只是工程师。 Internal tools can be built by designers with AI assistance, not just engineers.