Case study

Rhythmguard

Design-token and spacing-scale governance for CSS and Tailwind: enforce scales, require tokens and autofix arbitrary values before they ship.

toolingOpen sourceStylelintDesign tokensTailwindDesignOps

Live site

Estimated reading time

Key metrics

Autofix on production site

scale violations after one --fix pass, verified by re-lint

31 -> 0

use-scale benchmark

mean on 24k declarations vs 487ms for stylelint-scales

383ms

Test cases

node:test, plus 39 E2E assertions in a separate harness

119

Outcome

Three signals of impact.

Four Stylelint rules, two ESLint Tailwind rules, eight shareable configs and 18 researched scale presets, published to npm across 18 releases (v0.1.0 to v2.0.1)
Production migration measured on a live Next.js site: 31 scale violations autofixed to zero, verified by re-lint, with a clean 26-line substitution diff
Benchmarked against the nearest comparable rule on a 24,000-declaration corpus: use-scale mean 383ms vs 487ms, identical warning output

Role and scope

Creator and maintainer: rules, CLI, audit API, presets research and release engineering

Constraints

Zero test-framework dependencies: the suite runs on node:test, with a separate E2E fixtures repo proving the integration surface on real projects
Autofix determinism verified by running every fix scenario twice and asserting identical output against committed snapshots
Defaults are evidence-gated: the motion rule stays opt-in because a four-codebase audit showed the noise would outweigh the signal

Top decisions

Baseline and changed-only modes so legacy codebases can adopt by blocking only new drift
Audit JSON stabilized as a versioned contract with the old shape kept reachable
Scale presets derived from cited sources (Material, Atlassian, Carbon, baseline-grid literature), never invented

Confidential details redacted where required.

Next step

If you want this outcome, we can scope the engagement quickly.

Contact Work with me