I’m a working product designer, and the slowest part of every project is turning a rough wireframe into something consistent enough to hand to the front-end team. These three prompts do that heavy lifting for you, one layer at a time.
[Claude]
[Claude] Act as a senior design systems lead. Given the wireframe below for a SaaS dashboard (a sidebar, a stats row, a data table, and a modal), produce a one-page token spec in markdown: color palette with hex values and WCAG contrast notes, a 4-point type scale, spacing units, border radii, and elevation shadows. Map every token to a named role (surface, text, accent) not to a screen. Output as a table I can paste into Notion.
[ChatGPT]
[ChatGPT] You are a UI copy editor. Here is the raw button, empty-state, and error text from my checkout flow. Rewrite each string to match a calm, confident brand voice: under 12 words, one action per line, no passive voice. Then flag every string that could confuse a first-time user and suggest a shorter alternative. Deliver as a table with columns: location, current, revised, why.
[Cursor]
[Cursor] In this repo, there is a components folder with a Button, Card, and Input written in TypeScript and styled with Tailwind. Refactor them to consume design tokens from a single tokens.ts file instead of hardcoded values. Preserve the existing props API and add JSDoc types for each token. Then run the type check and fix any errors. Output a diff summary of files changed and token names introduced.
More daily packs are waiting in the archive at aitoolalliance.com/tag/prompt-pack.