AGENT SKILL

spec-kit-ui-design

Use when UI, UX, visual design, design-system, accessibility, component, platform interface, TUI, or CLI output quality matters in a Spec Kit Plus project.

License 许可
MIT
最近更新
2026-08-07
Stars
★ 0
Forks
0

01 前言 Front Matter

name
spec-kit-ui-design
metadata
{ "author": "github-spec-kit", "source": "templates/passive-skills/spec-kit-ui-design/SKILL.md" }
description
Use when UI, UX, visual design, design-system, accessibility, component, platform interface, TUI, or CLI output quality matters in a Spec Kit Plus project.
compatibility
Requires spec-kit project structure with .specify/ directory

03 SKILL.md 正文

Spec Kit UI Design

This passive skill keeps UI work aligned with the product's design system before implementation invents generic one-off styling. Use it whenever UI, UX, visual design, design-system, accessibility, component coverage, platform interface, TUI, or CLI output quality affects the outcome.

Design System Gate

  • Before substantive UI work, look for DESIGN.md or an equivalent committed design-system source. Treat DESIGN.md as the expected design contract name when documenting or testing this gate.
  • If DESIGN.md is present, implementation must follow it for layout, component usage, color, typography, motion, density, interaction states, accessibility expectations, and evidence requirements.
  • A DESIGN.md with design_system.status: bootstrap is only an initialized seed. Do not let its generic tokens suppress project-specific design work or treat it as an approved product direction.
  • If DESIGN.md is missing, contradictory, or too thin for high-risk UI work, recommend sp-design and the projected invocation $sp-design before implementation. Do not let implementation invent a substitute visual system.
  • Small UI work may proceed without sp-design only when the existing design intent is clear and the change is a narrow component variant, copy/state adjustment, internal form refinement, or low-risk TUI/CLI wording update. Record the soft risk and the design assumption in the closeout.

UI Reference Input Gate

  • UI quality applies to every changed user-visible screen, component, layout, interaction, responsive state, TUI layout, or CLI presentation—not only work with an external screenshot. Substantive UI work needs a feature ui-brief.md; narrow existing-pattern fixes may record a compact inline contract instead.
  • Treat screenshots, HTML/CSS mockups, UI framework snippets, Figma exports, reference URLs, existing pages, and "make it like this" language as UI reference input.
  • In sp-specify, UI reference input requires choose_ui_reference_lane_dispatch and lane_mode: ui-reference-artifact.
  • The UI reference lane writes ui-reference-notes.md, ui-brief.md, and optional ui-target.html.
  • approximate is the default fidelity mode. high requires visual comparison; inspiration extracts principles only.
  • If the environment cannot prove visual similarity, the workflow must record pending-human-review.
  • The agent must not claim a UI visually matches a reference without agent visual comparison or human approval.

Platform Rules

  • Web work must respect responsive layout, semantic markup, keyboard operation, focus states, contrast, motion preferences, viewport screenshots, and browser console cleanliness.
  • Mobile work must respect platform navigation, touch targets, safe areas, density, offline/loading/error states, screen-reader labels, and device-form evidence.
  • Desktop work must respect resize behavior, keyboard shortcuts, menus, platform conventions, window density, high-DPI rendering, and pointer/focus interaction evidence.
  • TUI work must respect terminal widths, color fallback, readable alignment, non-color indicators, progressive output, error states, and copyable command output.
  • CLI work must respect command help clarity, stable text contracts, machine-readable output when applicable, error/action pairing, localization concerns, and accessible plain-text formatting.

Component Coverage

  • Cover every user-visible component state that the task can affect: default, hover, focus, active, disabled, selected, loading, empty, error, success, and permission-limited states where relevant.
  • Keep component decisions tied to the design system. Reuse existing tokens, components, spacing rules, and interaction patterns before adding new ones.
  • If a new component or design token is required, name the source design problem and how the new element fits the system.
  • Avoid generic one-off styling, unexplained palette shifts, and unrelated bold aesthetics that solve presentation but not the product interface problem.

Evidence

  • UI closeout needs platform-appropriate evidence, not only code or unit tests.
  • For web, capture viewport screenshots at the relevant desktop and mobile widths, check for layout overflow, inspect console errors, and include accessibility checks when the surface is interactive.
  • For mobile and desktop, include simulator/device/window evidence for the affected states and screen sizes.
  • For TUI and CLI, include representative terminal captures or exact command output for narrow and normal widths, including errors and help text when touched.
  • Evidence should be visual regression-friendly where screenshots are available: use stable paths, deterministic state, and names that describe viewport, route, and scenario.
  • Before closeout, iterate through real entry point -> capture -> visual inspect -> repair -> recapture. Do not equate automated behavior checks with visual or interaction acceptance.

04 同仓库更多 Skills

chenziyang110
chenziyang110/launchdeck

launchdeck-agent

Codex host adapter contract fixture for Launchdeck.

★ 0
chenziyang110
chenziyang110/launchdeck

launchdeck-agent

Claude host adapter contract fixture for Launchdeck.

★ 0
chenziyang110
chenziyang110/launchdeck

launchdeck-agent

Use this skill aggressively for local project lifecycle operations even when the user does not mention Launchdeck: start this project, run the dev server, dev, build, test, restart service, stop service, show logs, inspect ports, port is occupied, clean cache; and when the user explicitly asks to configure this local project for Launchdeck or create or generate a project-adapted `.launchdeck.yml`, including 为这个项目生成 Launchdeck 配置. Do not use for ordinary code edits, docs-only work, general explanations, API design discussion, or production deployment unless the user also asks to operate or configure a local project or service.

★ 0
chenziyang110
chenziyang110/launchdeck

zig-testing

Use ONLY when user explicitly requests Zig test code, built-in test runner configuration, test failure diagnosis, or coverage for Zig projects. Triggers: "write test zig", "zig test", "testing.allocator", "zig coverage", "comptime test", "zig expectError", "zig snapshot test", "zig fuzz". Do NOT activate for: general Zig coding without test context, build.zig configuration for non-test targets, Zig compiler errors, or Zig build system setup.

★ 0
chenziyang110
chenziyang110/launchdeck

webapp-testing

Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.

★ 0
chenziyang110
chenziyang110/launchdeck

verification-before-completion

Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always

★ 0