suriyel/longtaskforagent/long-task-ucd
Claude Code and agent skill by suriyel/longtaskforagent
- community
Checking that the SKILL.md file is still in the repository.
Use when SRS doc exists but no UCD doc and no design doc and no feature-list.json - generate UI Component Design style guide with text-to-image prompts based on approved SRS
Install
npx vskill@latest install suriyel/longtaskforagent/long-task-ucdSKILL.md
UI Component Design (UCD) Style Guide Generation
Take the approved SRS as input. Analyze UI-related requirements, define visual style direction, and produce a UCD style guide containing text-to-image model prompts — so that all frontend features share a unified visual language.
<HARD-GATE> Do NOT invoke any design skill, implementation skill, write any code, scaffold any project, or take any implementation action until you have presented the UCD style guide and the user has approved it. This applies to EVERY project with UI features. </HARD-GATE>
When This Phase Applies
This phase runs after SRS approval and before design. It applies when:
- The approved SRS contains UI-related functional requirements (FR-xxx with user-facing screens, pages, or components)
- No UCD document (
*-ucd.md) exists indocs/plans/
If the SRS has NO UI features: announce "No UI features detected in SRS — skipping UCD phase" and immediately chain to design via long-task:long-task-design.
Checklist
You MUST create a TodoWrite task for each of these items and complete them in order:
- Read the approved SRS — from
docs/plans/*-srs.md - Extract UI scope — identify all UI-related requirements and user personas
- Define visual style direction — propose 2-3 style options with mood boards
- Generate component-level prompts — text-to-image prompts for each UI component type
- Generate page-level prompts — text-to-image prompts for each key page/screen
- Define style tokens — color palette, typography, spacing, iconography
- Present & approve UCD — section-by-section for non-trivial projects
- Save UCD document —
docs/plans/YYYY-MM-DD-<topic>-ucd.mdand commit - Transition to design — REQUIRED SUB-SKILL: Invoke
long-task:long-task-design
The terminal state is invoking long-task-design. Do NOT invoke any other skill.
Step 1: Read SRS & Extract UI Scope
- Read the approved SRS document from
docs/plans/*-srs.md - Extract UI-relevant inputs:
- User personas — technical level, accessibility needs, device preferences
- Functional requirements with UI — screens, pages, forms, dashboards, data visualizations
- NFR usability requirements — accessibility standards (WCAG level), responsive breakpoints, internationalization
- Constraints — brand guidelines, platform restrictions, browser support
- Interface requirements — external UI components, design systems to integrate with
- Build a UI inventory — list every distinct screen/page/component type implied by the SRS
- If the SRS lacks sufficient UI detail → ask user via
AskUserQuestionbefore proceeding
Step 2: Define Visual Style Direction
Present 2-3 visual style options to the user:
Wait for user to choose or provide direction. Incorporate feedback before proceeding.
Step 3: Generate Style Tokens
Define the concrete design tokens that anchor the entire style system:
3.1 Color Palette
- All contrast ratios MUST meet WCAG AA at minimum (4.5:1 for normal text, 3:1 for large text)
- If SRS specifies WCAG AAA, ratios must be 7:1 / 4.5:1
3.2 Typography Scale
3.3 Spacing & Layout
3.4 Iconography & Imagery
Step 4: Generate Component-Level Prompts
For each UI component type in the inventory, produce a text-to-image prompt that a generative image model (Midjourney, DALL-E, Stable Diffusion, etc.) can use to visualize the component.
Prompt Structure
Each component prompt follows this template:
Required Component Types
Generate prompts for at least these component types (skip only if truly absent from the UI inventory):
| Category | Components | |
|---|---|---|
| Navigation | Header/navbar, sidebar, breadcrumb, tabs, pagination | |
| Input | Text input, textarea, select/dropdown, checkbox, radio, toggle, date picker | |
| Action | Primary button, secondary button, icon button, link button, FAB | |
| Feedback | Alert/toast, modal/dialog, progress bar, skeleton loader, empty state | |
| Data Display | Table, card, list item, badge/tag, avatar, tooltip | |
| Layout | Page shell, form layout, grid/masonry, divider |
Step 5: Generate Page-Level Prompts
For each key page/screen identified in the UI inventory, produce a full-page text-to-image prompt.
Page Prompt Structure
Step 6: Present & Approve UCD
For non-trivial projects, present section by section:
- Visual style direction — mood, color tendency, typography direction
- Style tokens — color palette, typography scale, spacing, iconography
- Component prompts — one or two representative components for approval before generating the rest
- Page prompts — key pages for approval
Present each section. Wait for user feedback. Incorporate changes before moving to the next.
For simple projects (< 3 UI pages): combine all sections into a single approval step.
Step 7: Save UCD Document
Save the approved UCD style guide to docs/plans/YYYY-MM-DD-<topic>-ucd.md.
Document structure:
Step 8: Transition to Design
Once the UCD document is saved and committed:
- Summarize key inputs the design phase will need:
- From SRS: functional requirements, NFRs, constraints
- From UCD: style tokens, component catalog, page layouts → informs UI/UX section and frontend architecture in design doc
- REQUIRED SUB-SKILL: Invoke
long-task:long-task-designto begin design
Scaling the UCD Phase
| Project Size | UI Pages | Depth | |
|---|---|---|---|
| Tiny | 1-3 | Style tokens + 3-5 core component prompts + page prompts; single approval step | |
| Small | 3-8 | Full style tokens + component prompts for used components + all page prompts | |
| Medium | 8-20 | Full UCD with all component variants + responsive page prompts | |
| Large | 20+ | Full UCD + interaction state matrices + animation spec + dark mode variants |
Red Flags
This is the start of the file. Read the full SKILL.md on GitHub.
Security
- Security scanPassed, 100 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/long-task-ucd/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/suriyel/longtaskforagent/long-task-ucd)