agentleFS
cenconq25cenconq25/claude-code-app-studio/.claude/skills/team-design/SKILL.md

team-design

Skill40 starsChanged 7 months agoMITView on GitHub

Orchestrate a full design cycle. Coordinates lead-designer, ux-designer, visual-design-director, interaction-designer, and motion-designer to take a feature from brief through to implementation-ready specs.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No argument -> use AskUserQuestion to capture: feature name, target user, business goal, scope boundaries, deadline, prior art.

    line 45
  2. Very specific:

    If the proposal extends the system, propose updates to .claude/docs/design-system.md for review.

    line 101
  3. Very specific:

    Ask before writing to design/packages/[feature]-design.md.

    line 194
  4. Distinctive:

    Use AskUserQuestion to confirm scope.

    line 206
  5. Distinctive:

    If any subagent returns BLOCKED:

    line 226

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Team Design

End-to-end design cycle orchestration. Spawns the five design roles in the order their work depends on each other, gathers approvals at every phase boundary, and lands on a package the engineering team can implement without re-asking.


Team Composition

  • lead-designer — design direction, scope boundaries, sign-off authority.
  • ux-designer — user flows, screens, information architecture, accessibility.
  • visual-design-director — visual system, typography, color, imagery.
  • interaction-designer — gestures, affordances, microinteractions, states.
  • motion-designer — transitions, animations, haptics choreography.

Spawn each via Task. Run independent subagents in parallel where inputs are independent.


Phase 1: Brief Intake

Parse the argument:

  • --brief=<path> -> read the explicit brief.
  • --feature=<name> -> glob design/prd/**/*[name]*.prd.md and related stories.
  • No argument -> use AskUserQuestion to capture: feature name, target user, business goal, scope boundaries, deadline, prior art.

Read in parallel:

  • The PRD if present.
  • .claude/docs/design-system.md (if present) — current visual system.
  • Any existing flows for the feature in design/flows/.

Phase 2: Direction via lead-designer

Spawn lead-designer via Task. Prompt template:

Brief: [content]. Existing system: [paths]. Set design direction: the user-experience hypothesis, the success criteria for this design (number of taps, error rate, conversion), the explicit scope of this cycle (what is in, what is out), the design tier (custom one-off vs. system-conforming).

Render the direction. Use AskUserQuestion:

  • [A] Approve direction — proceed
  • [B] Adjust scope
  • [C] Reset — re-discuss the brief

Phase 3: User Flow via ux-designer

Spawn ux-designer via Task. Prompt template:

Direction: [reference]. Brief: [reference]. Produce: complete user flow diagram (entry points, main path, branches, error paths, happy/unhappy ends). Information architecture per screen (what's shown, hierarchy). Accessibility considerations: target dynamic type sizes, contrast, screen reader path, reduced-motion path, keyboard / external-keyboard support.

Render flow. Identify any open question for lead-designer to resolve. Loop on AskUserQuestion until approved.


Phase 4: Visual System via visual-design-director (parallel with Phase 5)

Spawn visual-design-director via Task. Prompt template:

Direction: [reference]. Flow: [reference]. Existing visual system: [paths]. Decide whether this feature uses the existing system verbatim, extends it (adds tokens), or breaks from it. For each decision, document the rationale. Output: typography, color, icon, imagery treatments. Specify dark mode and high-contrast variants.

If the proposal extends the system, propose updates to .claude/docs/design-system.md for review.


Phase 5: Interaction Spec via interaction-designer (parallel with Phase 4)

Spawn interaction-designer via Task. Prompt template:

Direction: [reference]. Flow: [reference]. For each screen and control, specify: idle state, pressed state, disabled state, error state, loading state, empty state. Specify gestures (tap, long-press, swipe, pinch, pull-to-refresh). Specify affordances and feedback (visual + haptic). Note platform conventions: iOS swipe-back, Android system back, share sheets, action sheets, contextual menus.

Render spec.


Phase 6: Motion via motion-designer

Spawn motion-designer via Task. Prompt template:

Direction: [reference]. Flow: [reference]. Interaction spec: [reference]. Visual system: [reference]. Choreograph transitions between screens, microinteractions for key controls, loading states, success / error animations. Specify duration and easing. Specify haptic pairings. Specify reduced-motion alternatives — every motion must have a reduced-motion fallback.

Render spec. Cross-check with interaction-designer's pressed-state specs to confirm motion durations don't fight the interaction spec.


Phase 7: Holistic Review via lead-designer

Spawn lead-designer via Task to run a final critique against:

  • Brief alignment.
  • Internal consistency across UX, visual, interaction, motion.
  • Existing pattern library reuse vs. novelty.
  • Accessibility (every flow must work in screen reader, dynamic type, reduced motion, high contrast).
  • Engineering implementability — flag anything that looks expensive to build (custom shaders, frame-budget-risky animations).

Render the verdict: APPROVED / NEEDS REVISION. If revision, loop the specific subagent for fixes.


Phase 8: Compose the Design Package

# Design Package — [feature]

## Direction
[from lead-designer]

## Flow
[from ux-designer — embedded as Mermaid or referenced as a Figma link]

## Visual System
[from visual-design-director — token table, dark/contrast variants]

## Interaction Spec
[from interaction-designer — per-control state matrix]

## Motion Spec
[from motion-designer — timing, easing, haptic pairings, reduced-motion]

## Accessibility
- Dynamic type range: [min..max]
- Contrast: WCAG AA / AAA
- Reduced motion fallbacks: [mapped]
- Screen reader path: [described]
- Reduced data: [described]

## Implementation Notes
- Frame-budget-risky moments: [list]
- Platform divergences: [list]
- Open questions for engineering: [list]

## Approvals
- Lead Designer
- UX Designer
- Visual Design Director
- Interaction Designer
- Motion Designer

Ask before writing to design/packages/[feature]-design.md.


Phase 9: Story Breakdown

Propose stories for /create-stories:

  • One per screen (UI type).
  • One per non-trivial animation (Visual/Feel type).
  • One per new visual-system token addition (Config/Data type).

Use AskUserQuestion to confirm scope.


Phase 10: Update State

Append to production/session-state/active.md:

## Design Package — [date]
- Feature: [name]
- Package: [path]
- Stories proposed: [count]
- Next: /ux-review then /create-stories then /dev-story

Error Recovery

If any subagent returns BLOCKED:

  • ux-designer blocked on missing PRD detail -> back to PRD author.
  • visual-design-director blocked on no existing system -> propose initial design system definition before continuing.
  • motion-designer blocked on platform-channel limitations (Flutter haptics on iOS, etc.) -> escalate to framework specialist.

Quality Gates / PASS-FAIL

  • PASS — every screen has flow + visual + interaction + motion spec; every motion has a reduced-motion fallback; all five approvals present; package implements every PRD AC.
  • FAIL — missing spec for any screen, or any AC not addressed in the package.

Examples

Example 1 — onboarding redesign: Brief: shorten onboarding from 6 screens to 3 without losing key information. Five subagents engaged sequentially with parallel V/I spec. Package contains: 3 screens, 8 microinteractions, 4 transitions, all reduced-motion fallbacks. 7 stories proposed.

Example 2 — single-screen polish: Lightweight cycle. ux-designer skipped (existing flow). Direction + visual + interaction + motion only. Package smaller; 2 stories proposed.


Next Steps

  • /ux-review for an additional review against UX standards.
  • /create-stories for the proposed stories.
  • /dev-story to begin implementation.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.