Request to add specific guidance to documentation on when to use Noto Serif for AI/generative prose content, distinguishing it from UI chrome fonts, as current documentation lacks specific instructions.
## Background Candor includes Noto Serif in its font stack but provides no guidance on when to use it. The existing documentation implies it's for article/long-form reading contexts. During the Sovereign Council migration, a more specific and reusable pattern emerged. ## The pattern: prose/UI font split In AI-augmented apps, there is a meaningful distinction between: - **UI chrome** — labels, buttons, badges, metadata, status messages. Written by developers. Should be in Roboto Flex / Atkinson Hyperlegible. - **AI-generated content** — response text, analysis, reasoning, minority positions. Written by a model. Should be visually distinguishable from UI text. Using Noto Serif for AI-generated prose creates a clear visual contract: **serif = the AI wrote this, sans = the UI is telling you something**. This distinction: - Helps users immediately identify which parts of the screen are model output vs. interface - Reduces the risk of confusing UI instructions with model-generated conte