Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Design Systems

ASecurity

Build UI inside an existing design system — find its tokens and components, reuse before adding, decide between extending a component and introducing a new one, and preserve conventions you did not set. Use when a project already has design tokens, a component library, a theme or a shared Figma library and you are about to add or change UI inside it. Not for establishing a visual direction where no system exists, and not for reviewing an interface's usability.

46 stars
0 votes
0 copies
0 views
Added 9/22/2026
ai-agentsexpressapifrontenddocumentation

Works with

api

Security Analysis

A100/100

Scanned 9/22/2026

Install to Claude Code

$npx -y skills add nahid-sparktales/agent-dispatcher --skill design-systems --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Design Systems?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Design Systems
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/nahid-sparktales-design-systems/badge)](https://www.skillsdirectory.com/skills/nahid-sparktales-design-systems)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: design-systems
description: Build UI inside an existing design system — find its tokens and components, reuse before adding, decide between extending a component and introducing a new one, and preserve conventions you did not set. Use when a project already has design tokens, a component library, a theme or a shared Figma library and you are about to add or change UI inside it. Not for establishing a visual direction where no system exists, and not for reviewing an interface's usability.
---

# Working inside a design system

A system's value is that everything in it agrees. The expensive mistakes are not ugly components —
they are the private token, the one-off override and the sixth button variant, each individually
reasonable, which together turn a system back into a pile of CSS.

## When this fires

The project already has tokens, a component library, a theme or a shared library, and you are
adding or changing UI inside it. It does not fire when there is no system and the decisions are
yours — that is `frontend-design`.

## Procedure

1. **Find the system before writing anything.** Locate the token source, the component directory,
   its documentation, and two or three real usage sites. Say where each lives. If the system is a
   third-party library, read its current documentation rather than recalling its API — versions
   move and a guessed prop is a silent fork.
2. **Read the conventions you did not set**: token naming and whether tokens are primitive or
   semantic, how variants are expressed (props, classes, compound components), how theming and
   dark mode resolve, file and export layout, and what the existing components already handle that
   you were about to reimplement.
3. **Take the highest rung that holds**, in this order:
   1. An existing component, as it is.
   2. An existing component with an existing variant or prop.
   3. Existing primitives composed together.
   4. An existing component extended with a new variant.
   5. A new component in the system.
   6. A local one-off, explicitly marked as not part of the system.

   Most work stops at rung 1 or 2. Reaching rung 5 for a single screen is almost always rung 3
   misread.
4. **Extend when the need is the same concept under a new condition** and the component's API stays
   coherent afterwards — a new size, a new tone of the same control. **Add when it is genuinely a
   different concept** and at least two or three real, existing uses want it. One speculative use
   is not a system component; build it locally and promote it when the second use arrives.
5. **Never hardcode a value a token covers.** If no token covers what you need, that gap is the
   finding: report it and choose the nearest token, or ask. Inventing a private token beside the
   system is how the system stops being one.
6. **Treat a shared component as every screen that uses it.** Before changing one, list its call
   sites. Changing a default, renaming or removing a prop, or altering a token's value is a change
   to all of them — stop and ask before doing it, and name the consumers in the question.
7. **Follow the convention even where you disagree.** Say your disagreement once, in the report,
   and then match what is there. A locally better choice that breaks the pattern costs more than
   it gains.
8. **Render what you built inside the real system**, not in isolation: every theme the system
   ships, both a small and a large viewport, and the component's own states. Token and variant
   changes are invisible in a diff and show up only on screen.
9. **Document the addition where the system documents things** — the same place, the same shape,
   with a usage example and the case it is *not* for. An undocumented component is one somebody
   reimplements next quarter.

## Checklist

- [ ] Token source, component directory and real usage sites located and named
- [ ] Existing components checked before anything new was written
- [ ] The chosen rung is named, and why the one above it did not hold
- [ ] No raw value used where a token exists; every gap reported rather than patched privately
- [ ] Call sites listed before any shared component or token was touched
- [ ] Shared-surface changes were asked about, not assumed
- [ ] Rendered in every theme and at two viewports
- [ ] New or extended components documented where the system documents things

## Failure handling

- **No system is findable** — say so before proceeding. Do not invent one mid-task; either the
  work is `frontend-design`, or the system exists somewhere you have not looked yet. Ask.
- **The system contradicts itself** — two patterns for the same thing. Follow the one in newer or
  more numerous use, say which you followed and that the conflict exists. Do not resolve it by
  adding a third.
- **The design calls for something the system cannot express.** Report the gap with the specific
  case. Do not override the system to fake it; an override is invisible to the next reader and
  survives longer than the reason for it.
- **A token change looks right on your screen.** It is not verified until rendered in each theme —
  a value that reads well on light can fail contrast on dark.
- **The system is owned by another team.** Proposing a change to it is outward-facing work: prepare
  it, stop, and let the user take it to them.

## Evidence to report

Where the system lives — token file, component path, documentation. Which rung you took and what
ruled out the one above. Every token and component reused, by name. Any gap found. For a shared
change: the call sites, and the approval you were given. Screenshots per theme and viewport. State
plainly what was reused versus created, and whether the result was rendered or only built —
composed from system parts is not the same as seen working.

Attribution

nahid-sparktalesnahid-sparktales
View sourceMore from nahid-sparktales →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode that cuts output tokens while keeping technical accuracy. Levels: lite, full, ultra and the wenyan variants. Use for /caveman, "caveman mode", "talk like caveman", "be brief" or "less tokens".

1066601 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3351 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

651 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →