DesignMD journal
Notes on designing with AI coding agents
How to write a DESIGN.md, express design tokens an LLM can actually use, and get agents like Claude Code and Cursor to build interfaces that look like your product.
Claude Code DESIGN.md: Consistent UI
A Claude Code review for DESIGN.md workflows: where repository context improves UI consistency, where the terminal agent needs help, and who it fits.
Read articleCodex DESIGN.md: Consistent UI
A practical Codex review for DESIGN.md workflows: pair AGENTS.md with visual rules, use repository context, and keep human visual review in the loop.
Read articleDashboard DESIGN.md for AI Agents
A behavior-first dashboard DESIGN.md for AI agents: define density, navigation, charts, tables, states, and accessibility so operational UI stays coherent.
Read articlev0 DESIGN.md: Consistent UI
A practical v0 review for DESIGN.md workflows: use fast visual exploration without losing product rules, reusable components, and a reliable review loop.
Read articleWhy AI Coding Agents Need DESIGN.md
AI agents can write working UI, but they cannot guess your product’s design decisions. DESIGN.md turns visual intent into a repeatable coding workflow.
Read articleCursor DESIGN.md: Consistent UI
A practical Cursor review for DESIGN.md workflows: how project rules improve UI consistency, why the editor loop helps, and where drift still appears.
Read articleHow to Write a DESIGN.md Agents Follow
A practical DESIGN.md structure for AI coding agents: what to include, how to phrase rules clearly, and which anti-patterns prevent generic UI.
Read articleDESIGN.md Design Tokens: A Guide
How to express color, type, spacing, and shape tokens in DESIGN.md so AI agents apply the right values consistently instead of guessing.
Read articleDESIGN.md in an AI Agent Workflow
A practical workflow for using DESIGN.md with Claude Code, Cursor, and Codex: load the file, review output, and turn repeated corrections into rules.
Read article