Design System Ops plugin
A Claude Code skill pack: token and component audits, governance, docs and reports for a design system team
- 219 ★ on the repository
- Last commit
- MIT
Minimal access
Why this level
- Reads your token, component and documentation files and writes reports, making edits only on your request
Install
- Works in
Manual install
git clone https://github.com/murphytrueman/design-system-ops.git ~/.claude/skills/design-system-opsCheck loading with claude plugin list; for a project place the folder in .claude/skills/ and trust the project.
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- Reads your token, component and documentation files and writes reports, making edits only on your request
This is third-party code. Review the repository files before installing.
What it does
Design System Ops is a skill pack for Claude Code and Cowork for people who run a design system. The skills read your actual token and component files, find tier leakage between token levels, naming violations and divergence from the design, and produce a finding table with priority, remediation advice and effort ranges. Skills are grouped by task: audit (tokens, components, system health, drift, naming, Figma variables, themes, docs coverage), governance (contribution process, deprecation, decision records, change communication, backlog, version advice, codemods), documentation (component descriptions for AI, patterns, tokens, usage guidelines, AGENTS.md), validation (design-to-code match, accessibility, token compliance, schemas, component API, CI/CD) and communication (adoption report, stakeholder brief, pitch, onboarding). Four commands chain the skills: full-diagnostic, release-check, governance-review and token-migration.
Who it is for. Design system leads, senior design engineers and anyone responsible for a production design system with its tokens, components and documentation.
Good fit when
- Tokens have sprawled and you need a check of levels, names and leakage between layers
- Before a component release you run design match, accessibility and token compliance checks
- A component is being deprecated and you need a migration plan with a codemod and a message to consumers
- A short update on the system's state is being prepared for leadership
Not a fit when
- You need mockups and interface design: the set works on the system itself and product screens are outside its scope
- You want drift between teams without consumer code: drift-detection needs a consuming codebase
- You need work outside Claude Code and Cowork: the skills and commands target them
Example request
Audit my tokens in tokens/ for leakage between levels and naming violations and show a finding table with prioritiesLimitations
The pack targets Claude Code and the Cowork app. A project-level install only loads once you trust the project folder in the Claude Code prompt. Drift checks need code that consumes your system, and the Figma variables audit needs access to the relevant data. Reports come as finding tables with priorities and effort ranges, and applying fixes stays with the team. The last commit in the repository dates from 24 September 2026.
How to disable. For Claude Code delete ~/.claude/skills/design-system-ops or run /plugin uninstall for the plugin. For Cowork remove the design-system-ops plugin in the installed list.
FAQ
Which commands does it include?
/design-system-ops:full-diagnostic (six audits with a cross-skill summary), release-check (pre-release checks for a component), governance-review (adoption report, drift and a stakeholder brief) and token-migration (audit, transformation table, codemods and a staged rollout).
How do I install it in Cowork?
Download design-system-ops.plugin from the installable folder or the latest release, open a Cowork session in the Claude desktop app, drop the file into the chat and follow the install prompt.
What are three-tier tokens?
A scheme of primitive, semantic and component tokens. token-audit looks for tier leakage and checks format alignment with DTCG 2025.10.
Does it help with AGENTS.md?
Yes. The agent-instructions skill writes the AGENTS.md that Claude, Cursor and Copilot read, with a source for every rule and links to the inventory, dependency graph and component metadata.
What frameworks does it use?
Three-tier tokens, the Component Challenge Rating (how hard a component is to implement), a five-stage design system maturity model and an AI-readiness assessment.
Discussion
No questions or reviews yet.
Related
Pluginsdesign.how: 6 skills turn an app's UI into a design system for the agent, plus four ready systems for shadcn.
A skill for Figma design systems: read-only audits with WCAG checks, component building and token export to code.
A Claude skill that turns sites, images and Figma files into a design system: Markdown, DTCG tokens and element rebuild prompts
A workspace for Claude Code: ten commands lead from a product idea to React screens and a handoff package with ready prompts for implementation
A skill and CLI that pull colors, type, spacing, radius and shadows from a public site and save them as JSON and CSS tokens
A design skills pack for agents: research, UX strategy, design systems, UI, interaction, prototyping and visual critique
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Design System Ops (this card) | no | 219 | ||
| Design Systems MCP | no | not required | 227 | gives an agent access to design system data, while Design System Ops audits and governs your own system |
| UI UX Pro Max | no | 148,899 | picks styles and palettes for interfaces, while this works with tokens, versions and system drift |
README in short
The README states the job: tools for people who run a design system. Then come installation for Cowork and Claude Code, skill tables by category, four chained commands, material for coding agents, the core frameworks and sample requests such as auditing tokens, with a which-skill-do-I-need table.