Design-token reconstruction
Infer the smallest token set that explains consistent patterns in a reference.
Customize the recipe
0/1500 characters
Prompt preview
# Design-token reconstruction Infer the smallest token set that explains consistent patterns in a reference. ## Deliverable An evidence-backed token inventory with reusable mappings and example usage. ## Stack and visual direction - Recommended starting stack: Use the upstream REA project for supported artifact analysis. Deliver findings as a Markdown report with evidence references. - Swiss grid: Use a visible modular grid, aligned baselines, asymmetric composition and tight typographic hierarchy. Reflow the grid without shrinking essential text. Use this direction for any prototype or visual deliverable without altering what the evidence shows. - Deliver the evidence-backed report and any diagrams or specifications requested by the recipe. ## Tools and setup · Universal Use the upstream Reverse Engineer Anything skill as the sole specialized analysis resource. Apply it to supported, authorized builds when the question needs artifact evidence. Scope the work to observable behavior and documented access. When available evidence is sufficient or the REA connection is unavailable, work from the supplied material and record whether REA was applicable and used. - [Reverse Engineer Anything Skill](https://github.com/morluto/rea/blob/main/.agents/skills/reverse-engineer-anything/SKILL.md) (Skill) — Use the upstream REA workflow when a conclusion needs evidence from a supported, authorized built or shipped app. Artifact analysis requires an existing, compatible REA connection. - [REA Repository](https://github.com/morluto/rea) (Repository) — Consult the upstream documentation for supported artifacts and evidence requirements. ## Workflow 1. Collect repeated spacing, typography, color and radius observations across supplied screens. Propose primitive and semantic mappings, record confidence, and flag outliers requiring review. 2. Use the simplest implementation that delivers the stated outcome. Follow the user's project constraints and treat the starting stack as a default. Reuse existing work and ask for essential input only when it would change the result. Use realistic labeled sample data when real inputs are unavailable. 3. Run the applicable checks and report what actually passed along with any checks you could not run. ## Acceptance checks - [ ] Token choices are supported by multiple observed examples or labeled as estimates. - [ ] Semantic color pairs are checked for readable contrast in any prototype. - [ ] Work from supplied material, public references you can inspect, or systems and code the user owns or has permission to analyze. Cite only observed locations or the ranges supplied. Label reconstructions, assumptions and unknowns explicitly, including unseen validation or authentication. - [ ] For a prototype, use original or licensed materials and verify the relevant responsive, keyboard and accessible interaction behavior.
Recipe resources
Source, setup, and compatibility details.
Reverse Engineer Anything Skill
Use the upstream REA workflow when a conclusion needs evidence from a supported, authorized built or shipped app.
Artifact analysis requires an existing, compatible REA connection.
License · MIT
REA Repository
Consult the upstream documentation for supported artifacts and evidence requirements.
License · MIT for REA-authored code