Update to latest version

This commit is contained in:
Mårten Angner 2026-01-29 08:11:32 +01:00
parent 9cd83242fa
commit 044c83eeda
52 changed files with 2355 additions and 982 deletions

View File

@ -31,7 +31,7 @@ The HTML to Design workflow focuses on:
### Key Components Built
#### 1. Dev Mode JavaScript Component
- **Location**: `src/modules/wds/workflows/4-ux-design/interactive-prototypes/templates/components/dev-mode.js`
- **Location**: `src/modules/wds/workflows/4-ux-design/agentic-development/templates/components/dev-mode.js`
- **Purpose**: Interactive tool for extracting Object IDs from live prototypes
- **Features**:
- Toggle dev mode with button or Ctrl+E
@ -42,7 +42,7 @@ The HTML to Design workflow focuses on:
- Form field protection (Shift disabled when typing)
#### 2. Work File Template
- **Location**: `src/modules/wds/workflows/4-ux-design/interactive-prototypes/templates/work-file-template.yaml`
- **Location**: `src/modules/wds/workflows/4-ux-design/agentic-development/templates/work-file-template.yaml`
- **Purpose**: Complete planning document for section-by-section implementation
- **Structure**:
- Metadata and device compatibility

View File

@ -13,7 +13,7 @@ This document summarizes the work done on the HTML to Design side of the WDS v6
- ESLint rule compliance (`unicorn/prefer-module`, `no-unused-vars`, etc.)
### 2. Dev Mode JavaScript Component
- **Location**: `src/modules/wds/workflows/4-ux-design/interactive-prototypes/templates/components/dev-mode.js`
- **Location**: `src/modules/wds/workflows/4-ux-design/agentic-development/templates/components/dev-mode.js`
- **Purpose**: Developer/feedback mode for copying Object IDs in prototypes
- **Features Implemented**:
- Toggle dev mode with button or Ctrl+E
@ -24,7 +24,7 @@ This document summarizes the work done on the HTML to Design side of the WDS v6
- Form field protection (Shift disabled when typing)
### 3. Work File Template
- **Location**: `src/modules/wds/workflows/4-ux-design/interactive-prototypes/templates/work-file-template.yaml`
- **Location**: `src/modules/wds/workflows/4-ux-design/agentic-development/templates/work-file-template.yaml`
- **Purpose**: Complete planning document for section-by-section implementation
- **Structure**:
- Metadata and device compatibility

View File

@ -118,9 +118,9 @@ Choose a workflow to start:
```
*Define your product vision and goals*
### 🎨 **Interactive Prototypes**
### 🎨 **Agentic Development**
```
@wds/workflows/interactive-prototypes
@wds/workflows/agentic-development
```
*Build clickable prototypes for testing*
@ -158,7 +158,7 @@ Choose a workflow to start:
|----------|---------|-----------|
| **Trigger Map** | User pain points | `@wds/workflows/trigger-map` |
| **Product Brief** | Product vision | `@wds/workflows/product-brief` |
| **Prototypes** | Interactive demos | `@wds/workflows/interactive-prototypes` |
| **Prototypes** | Interactive demos | `@wds/workflows/agentic-development` |
| **Scenario Init** | User journeys | `@wds/workflows/scenario-init` |
---

View File

@ -0,0 +1,193 @@
# Getting Things Done (GTD)
**By:** David Allen (2001)
**Source:** "Getting Things Done: The Art of Stress-Free Productivity"
---
## Core Concept
GTD is a personal productivity methodology that helps manage commitments, information, and action. The core insight: your mind is for having ideas, not holding them. Capture everything externally, then organize and act systematically.
---
## The Five Steps
### 1. Capture
Collect everything that has your attention into a trusted system. Don't rely on memory.
### 2. Clarify
Process what you've captured:
- Is it actionable?
- What's the next action?
- What's the desired outcome?
### 3. Organize
Put things where they belong:
- Next Actions (by context)
- Projects (multi-step outcomes)
- Waiting For (delegated)
- Someday/Maybe (future ideas)
- Reference (information)
### 4. Reflect
Review your system regularly:
- Daily: Check calendar, review next actions
- Weekly: Full review of all lists, projects, commitments
### 5. Engage
Do the work with confidence, knowing your system has captured everything.
---
## The 2-Minute Rule
**If an action takes less than 2 minutes, do it immediately.**
This is one of GTD's most powerful and widely-adopted principles:
- The overhead of tracking a 2-minute task exceeds doing it
- Immediate completion clears mental clutter
- Builds momentum through quick wins
- Prevents small tasks from accumulating
### In Practice
| Situation | Action |
|-----------|--------|
| Quick email reply | Send now |
| Simple file rename | Do now |
| Brief clarification | Ask now |
| Small fix | Fix now |
### The Threshold
2 minutes is a guideline, not a strict rule. The principle is:
> **Planning overhead should not exceed task complexity.**
If documenting, categorizing, and scheduling a task takes longer than doing it — just do it.
---
## Applied in WDS Agentic Development
### Task Complexity Assessment
Before adding something to the plan, ask:
| Question | If Yes → |
|----------|----------|
| Can I fix this in < 2 minutes? | Do it now, log as sub-step |
| Does it need context I don't have? | Add to plan |
| Does it affect architecture? | Add to plan |
| Is it outside current scope? | Level 4 change request |
### Bug Fixes vs. Features
**Bugs (2-minute candidates):**
- Missing condition check
- Wrong variable name
- Off-by-one error
- Missing translation
**Features (need planning):**
- New component
- New state handling
- Architectural changes
- Multi-file refactors
### Sub-Step Pattern
When a 2-minute fix arises during planned work:
1. **Do** the fix immediately
2. **Log** it as a sub-step (e.g., 20a-1)
3. **Continue** with the main task
This maintains traceability without planning overhead.
---
## Why This Matters for AI Collaboration
Agentic development involves constant micro-decisions:
- Should I plan this?
- Should I do this now?
- Should I defer this?
The 2-minute rule provides a clear heuristic:
```
IF task_complexity < 2_minutes:
execute_immediately()
log_as_substep()
ELSE:
add_to_plan()
```
This prevents:
- Analysis paralysis on trivial tasks
- Planning overhead exceeding task value
- Context switching from minor interruptions
- Accumulated technical debt from deferred tiny fixes
---
## Related Concepts
### Inbox Zero
Process everything to zero — don't leave items in limbo.
### Next Actions
Define the very next physical action for every project.
### Waiting For
Track delegated items so nothing falls through cracks.
### Weekly Review
Regular system maintenance keeps it trustworthy.
---
## Source Materials
### Book
- **"Getting Things Done: The Art of Stress-Free Productivity"** by David Allen (2001, revised 2015)
### Website
- [GettingThingsDone.com](https://gettingthingsdone.com)
### Key Articles
- "The 2-Minute Rule" — David Allen
- "GTD in 15 minutes" — Various summaries online
---
## WDS Integration Points
| WDS Context | GTD Application |
|-------------|-----------------|
| **Agentic Development** | 2-minute rule for bug fixes, sub-step logging |
| **Agent Dialogs** | Capture system for ideas (dialog files) |
| **Change Requests** | Someday/Maybe list for outside-scope items |
| **Progress Logs** | Review and reflect on completed work |
| **Session Start Protocol** | Weekly review concept (check reality vs. plan) |
---
## Quick Reference
### The 2-Minute Rule
> If it takes less than 2 minutes, do it now.
### The Core Principle
> Planning overhead should not exceed task complexity.
### Applied to Development
> Quick fixes → Do and log as sub-step
> Complex changes → Add to plan first
---
*Getting Things Done - Stress-free productivity through systematic capture and action.*

View File

@ -93,7 +93,7 @@ Used in Trigger Mapping Workshop 1 to transform vision (visionary statements) in
### [Kathy Sierra Badass Users](./kathy-sierra-badass-users.md)
**By:** Kathy Sierra (2000s-2015)
**By:** Kathy Sierra (2000s-2015)
**Core Idea:** Don't make a better product. Make users better at what they want to do. Focus on user capability.
**Applied in WDS:**
@ -106,6 +106,21 @@ Used in Trigger Mapping Workshop 1 to transform vision (visionary statements) in
---
### [Getting Things Done (GTD)](./gtd-getting-things-done.md)
**By:** David Allen (2001)
**Core Idea:** Your mind is for having ideas, not holding them. Capture everything, then process and act systematically.
**Applied in WDS:**
- **2-Minute Rule:** If it takes < 2 minutes, do it now (don't plan trivial tasks)
- Agent Dialogs as capture system (externalize all ideas)
- Change Requests as "Someday/Maybe" list
- Session Start Protocol as weekly review concept
**When to Use:** Deciding whether to plan or act, managing task complexity, avoiding planning overhead
---
## Models vs. Methods
### Models (This Folder)
@ -227,6 +242,7 @@ These models represent decades of insight from brilliant thinkers who've shaped
- **Mijo Balic & Ingrid Domingues** - Connecting goals to user behavior
- **Gojko Adzic** - Making strategic planning accessible
- **Simon Sinek** - Teaching us to start with WHY
- **David Allen** - Systematic productivity and the 2-minute rule
- **George T. Doran** - Creating measurable, achievable objectives
- **Cathy Moore** - Focusing on action over information
- **Kathy Sierra** - Championing user capability

View File

@ -59,9 +59,10 @@ agent:
**Micro-Guides (load when needed):**
- Strategic Design → data/agent-guides/freya/strategic-design.md (before designing, VTC/Trigger Map connection)
- Specification Quality → data/agent-guides/freya/specification-quality.md (creating specs, logical explanations)
- Interactive Prototyping → data/agent-guides/freya/interactive-prototyping.md (HTML prototypes, validation)
- Agentic Development → data/agent-guides/freya/agentic-development.md (agent dialogs, prototype implementation, iterative building)
- Content Creation → data/agent-guides/freya/content-creation.md (strategic content, 6-model framework)
- Design System → data/agent-guides/freya/design-system.md (Phase 5, organic growth, Figma integration)
- Stitch Generation → workflows/4-ux-design/stitch-generation/workflow.md (AI-assisted UI generation with Google Stitch)
**Collaboration:**
- My domain: Phases 4 (UX Design), 5 (Design System - optional), 7 (Testing)
@ -74,6 +75,7 @@ agent:
- Prototypes validate before production (micro-guide: interactive-prototyping.md)
- Content is strategic, not decorative (micro-guide: content-creation.md)
- Design systems grow organically (micro-guide: design-system.md if Phase 5)
- AI-assisted design via Google Stitch when spec + sketch ready (workflow: stitch-generation)
- Visual refinement via Figma when design system incomplete (automated MCP integration)
**Project Tracking:**
@ -88,12 +90,20 @@ agent:
- trigger: ux-design
exec: "{project-root}/{bmad_folder}/wds/workflows/4-ux-design/workflow.md"
description: Create interactive prototypes and scenarios (Phase 4)
description: Create specifications and scenarios (Phase 4)
- trigger: agentic-development
exec: "{project-root}/{bmad_folder}/wds/workflows/9-agent-dialogs/workflow.md"
description: Build features iteratively with agent dialogs (prototypes, implementations, bug fixes)
- trigger: audit-spec
exec: "{project-root}/{bmad_folder}/wds/workflows/4-ux-design/specification-audit-workflow.md"
description: "[AS] Audit page or scenario specifications for completeness and quality"
- trigger: stitch-generation
exec: "{project-root}/{bmad_folder}/wds/workflows/4-ux-design/stitch-generation/workflow.md"
description: Generate UI designs with Google Stitch AI from specifications and sketches
- trigger: design-system
exec: "{project-root}/{bmad_folder}/wds/workflows/5-design-system/workflow.md"
description: Build component library with design tokens (Phase 5 - optional)

View File

@ -0,0 +1,200 @@
# Freya's Agentic Development Guide
**When to load:** When implementing features, building prototypes, or fixing bugs through structured agent dialogs
---
## Core Principle
**Agentic Development uses structured dialogs to build incrementally with full traceability.**
Agent Dialogs bridge the gap between specifications and working code. Each step is self-contained, allowing fresh context while maintaining continuity.
---
## What is Agentic Development?
Agentic Development is a **workflow approach** that produces various outputs:
| Output Type | Description | When to Use |
|-------------|-------------|-------------|
| **Interactive Prototypes** | HTML prototypes that let users FEEL the design | Validating UX before production |
| **Prototype Implementation** | Building features from specifications | Feature development |
| **Bug Fixes** | Structured debugging and fixing | Issue resolution |
| **Design Exploration** | Exploring visual/UX directions | Creative iteration |
**Key Insight:** By structuring work into documented dialog folders, we create:
- **Isolation** — Each step can run in a fresh context
- **Traceability** — Clear record of what was planned and executed
- **Replayability** — Instructions can be rerun if needed
- **Handoff** — Different agents or humans can continue the work
---
## Agent Startup Protocol
**When awakened, always check for pending dialogs:**
```
1. Check: docs/F-Agent-Dialogs/
2. Find dialogs where:
- Status = "Not Started" or "In Progress"
- Agent matches the awakened agent
3. Present pending dialogs to user
```
This ensures no captured work is forgotten.
---
## The Bridge Role
Agent Dialogs bridge **specifications** and **development**:
```
┌─────────────────────┐ ┌─────────────────────┐ ┌─────────────────────┐
│ SPECIFICATION │ │ AGENT DIALOG │ │ DEVELOPMENT │
│ │ │ │ │ │
│ • What to build │────────▶│ • What's in scope │────────▶│ • How to build │
│ • Object IDs │ │ • Step breakdown │ │ • Code files │
│ • Requirements │ │ • Traceability │ │ • Components │
│ • Translations │ │ • Progress tracking │ │ • Tests │
└─────────────────────┘ └─────────────────────┘ └─────────────────────┘
Single Source Navigation Implementation
of Truth Layer
```
**The specification is the single source of truth.** Dialogs do not duplicate spec content — they map implementation tasks to spec sections via Object IDs.
---
## Dialog Folder Structure
```
docs/F-Agent-Dialogs/
└── {DATE}-{agent}-{feature-name}/
├── {DATE}-{agent}-{feature-name}-dialog.md ← Main file
└── steps/
├── 01-{step-name}.md ← Self-contained steps
├── 02-{step-name}.md
└── ...
```
---
## Feedback Protocol
During implementation, classify and handle feedback naturally:
| Type | What It Is | When to Address |
|------|------------|-----------------|
| **Bug/Issue** | Something broken or not working as expected | Now — iterate until fixed |
| **Quick Adjustment** | Small tweak to current work | Now — implement immediately |
| **Addition** | New requirement that fits current scope | Later step — add to plan |
| **Change Request** | Outside current dialog scope | Future session — document in Change Requests |
**Response Pattern:**
1. **Classify** — Note what kind of feedback this is
2. **Timing** — State when it should be addressed
3. **Confirm** — For additions and change requests, confirm before proceeding
4. **Execute** — Implement or document as appropriate
---
## Interactive Prototypes (Output Type)
Interactive Prototypes are **one output** of Agentic Development.
### Why HTML Prototypes?
**Static Specs Can't Show:**
- How it FEELS to interact
- Where users get confused
- What's missing in the flow
- If the pacing feels right
**HTML Prototypes Reveal:**
- Interaction feels natural or awkward
- Information appears when needed
- Flow has logical gaps
- Users understand next steps
### Fidelity Levels
| Level | Focus | Use When |
|-------|-------|----------|
| **Wireframe** | Information architecture | Testing flow logic only |
| **Interactive** | User experience | Validating UX (standard) |
| **Design System** | Component-based | Phase 5 enabled |
### Prototype vs Production
**Prototypes ARE:**
- Thinking tools
- Communication tools
- Validation tools
- Specification supplements
**Prototypes are NOT:**
- Production code
- Pixel-perfect mockups
- Final design
---
## Prototype Implementation (Output Type)
Building features from specifications through structured dialog steps.
### Step File Structure
Each step links to specifications (doesn't duplicate):
```markdown
## Object ID Implementation Map
| Object ID | Spec Section | Lines |
|-----------|--------------|-------|
| `booking-detail-header` | Drawer Header | L149-L158 |
| `booking-detail-close` | Close Button | L159-L168 |
```
### Implementation Checklist Pattern
For each Object ID:
1. **Read** — Load the spec section
2. **Implement** — Build to match spec
3. **Verify** — Confirm Object ID present and behavior correct
---
## Best Practices
### Single Source of Truth
- **Never duplicate spec content** — Link to spec sections with line numbers
- **Object IDs are the contract** — Every implementation maps to an Object ID
- **Spec changes update the spec** — Not the dialog or step files
### Dialog Files
- **Be thorough in Setup Context** — Assume zero prior knowledge
- **Include file paths** — Always use absolute or project-relative paths
- **Track progress** — Update the Steps Overview table after each step
### Execution
- **Read spec first** — Before implementing any Object ID
- **Fresh context is fine** — Steps are designed to work in isolation
- **Update as you go** — Don't wait to update progress
- **Capture discoveries** — Note spec changes or issues found
---
## Related Resources
- **Agent Dialog Workflow:** `workflows/9-agent-dialogs/workflow.md`
- **Dialog Template:** `workflows/9-agent-dialogs/templates/dialog.template.md`
- **Step Template:** `workflows/9-agent-dialogs/templates/step.template.md`
- **Phase 4 UX Design:** `workflows/4-ux-design/workflow.md`
---
*Build incrementally. Document thoroughly. Let users FEEL the design before committing to production.*

View File

@ -1,259 +0,0 @@
# Freya's Interactive Prototyping Guide
**When to load:** When creating HTML prototypes or interactive mockups
---
## Core Principle
**HTML prototypes are THINKING TOOLS, not final products.**
Prototypes let users FEEL the design before we commit to production code. They reveal gaps in logic that static specs might miss.
---
## Why HTML Prototypes?
### Static Specs Can't Show
- ❌ How it FEELS to interact
- ❌ Where users get confused
- ❌ What's missing in the flow
- ❌ If the pacing feels right
- ❌ Whether copy actually works
### HTML Prototypes Reveal
- ✅ Interaction feels natural or awkward
- ✅ Information appears when needed
- ✅ Flow has logical gaps
- ✅ Users understand next steps
- ✅ Content triggers emotions
---
## Prototypes as Validation
**Think of prototypes as:**
- **Thinking tools** - Help us understand what we're building
- **Communication tools** - Show stakeholders/users the vision
- **Validation tools** - Catch problems before coding
- **Specification supplements** - Demonstrate what words can't
**NOT:**
- ❌ Production code (that's Phase 4 → BMM handoff)
- ❌ Pixel-perfect mockups (that's Figma's job)
- ❌ Final design (they're meant to evolve)
---
## When to Create Prototypes
### Always Create For
- Complex interactions (multi-step forms, wizards)
- Novel UI patterns (users haven't seen before)
- Critical flows (signup, purchase, onboarding)
- Content-heavy pages (validate information hierarchy)
### Optional For
- Simple pages (standard layouts)
- Repetitive patterns (once validated, reuse)
- Admin interfaces (if similar to known patterns)
**When in doubt → Prototype.** 30 minutes of HTML saves hours of rework.
---
## Prototype Fidelity Levels
### 1. Wireframe Prototype (Fastest)
- Basic HTML structure
- Placeholder content
- No styling (browser defaults)
- Focus: Information architecture
**Use when:** Testing flow logic only
---
### 2. Interactive Prototype (Standard)
- Structured HTML
- Actual content (multi-language)
- Basic CSS (layout, spacing, typography)
- Interactive elements (buttons, forms, navigation)
**Use when:** Validating user experience
---
### 3. Design System Prototype (If Enabled)
- Component-based HTML
- Design System classes
- Design tokens (colors, spacing)
- Real interactions
**Use when:** Phase 5 (Design System) is enabled
---
## Using Design System Components
### If Design System Enabled (Phase 5)
**Check first:**
1. Does this component exist in the Design System?
2. If yes → Use it
3. If similar → Assess opportunity/risk of creating variant
4. If no → Mark for future extraction
**In prototype:**
```html
<!-- Reference existing component -->
<button class="ds-button ds-button--primary">
Sign Up
</button>
<!-- Or note for future -->
<!-- TODO: Extract as ds-card-feature component -->
<div class="temp-feature-card">
...
</div>
```
---
### If Design System Disabled
**Use page-specific classes:**
```html
<!-- Page-specific, won't be reused -->
<button class="landing-cta-primary">
Sign Up
</button>
```
**Developers know:** No design system = implement as-is, no abstraction needed.
---
## What to Include in Prototypes
### Must Have
- ✅ All sections in correct order
- ✅ Actual content (headlines, copy, CTAs)
- ✅ Multi-language versions (separate HTML files or language toggle)
- ✅ Interactive elements (buttons, forms, links)
- ✅ Key states (default, hover, active, disabled)
### Optional
- Form validation (unless testing UX)
- Backend integration (never)
- Pixel-perfect design (Figma's job)
- Production-quality code (it's a prototype!)
---
## Prototype Validation Process
### 1. Internal Check
- Click through the flow yourself
- Does it feel natural?
- Any confusing moments?
- Missing information?
- Logical gaps?
### 2. Stakeholder Review
- Show prototype in conversation
- Watch where they pause or ask questions
- Note confusion points
- Validate assumptions
### 3. User Testing (Optional)
- If critical flow, test with 3-5 users
- Watch, don't explain
- Note where they struggle
- Identify patterns
### 4. Iterate
- Fix gaps revealed by validation
- Update specification accordingly
- Re-prototype if major changes
---
## Prototype → Specification Flow
**Prototypes inform specs, not replace them:**
1. **Create prototype** - Think through interaction
2. **Validate prototype** - Catch issues early
3. **Update specification** - Document what works
4. **Generate final spec** - With prototype insights
**Result:** Specification is battle-tested before development.
---
## Common Prototype Mistakes
### ❌ Over-Engineering
"Let me make this perfect production code..."
- **Why bad:** Wastes time, misses the point
- **Instead:** Quick and dirty is fine - it's a prototype
### ❌ Under-Engineering
"Just some divs with text..."
- **Why bad:** Can't validate actual experience
- **Instead:** Make it interactive enough to feel real
### ❌ Skipping Validation
"I know this works, no need to test..."
- **Why bad:** Your assumptions might be wrong
- **Instead:** Always validate with at least one other person
### ❌ Treating as Final
"This prototype IS the spec..."
- **Why bad:** Missing critical specification details
- **Instead:** Prototype → insights → specification
---
## Technical Notes
### File Structure
```
docs/C-Scenarios/[scenario-name]/prototypes/
├── landing-page-en.html
├── landing-page-se.html
├── signup-flow-en.html
└── styles.css (shared)
```
### Basic Template
```html
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>[Page Name] - Prototype</title>
<link rel="stylesheet" href="styles.css">
</head>
<body>
<!-- Prototype content -->
<!-- Note: This is a UX prototype, not production code -->
</body>
</html>
```
---
## Related Resources
- **Phase 4 Workflow:** `../../workflows/4-ux-design/`
- **Design System:** `../../workflows/5-design-system/`
- **Page Specification Template:** `../../workflows/4-ux-design/templates/page-specification.template.md`
---
*Show, don't tell. Let users FEEL the design before we build it.*

View File

@ -28,7 +28,7 @@
## ✅ DO Use Specific Names
### Always Create:
- ✅ `INTERACTIVE-PROTOTYPES-GUIDE.md` (specific topic)
- ✅ `AGENTIC-DEVELOPMENT-GUIDE.md` (specific topic)
- ✅ `FREYA-WORKFLOW-INSTRUCTIONS.md` (specific agent + purpose)
- ✅ `PROTOTYPE-ROADMAP.md` (specific purpose)
- ✅ `PROJECT-ANALYSIS-ROUTER.md` (specific function)
@ -43,7 +43,7 @@
### Pattern 1: [TOPIC]-GUIDE.md
**When**: Overview/introduction to a topic
**Examples**:
- `INTERACTIVE-PROTOTYPES-GUIDE.md`
- `AGENTIC-DEVELOPMENT-GUIDE.md`
- `DESIGN-SYSTEM-GUIDE.md`
- `TESTING-GUIDE.md`
@ -199,7 +199,7 @@ project-analysis/
- `-ROADMAP.md` - Status/plan tracking
**Examples**:
- `INTERACTIVE-PROTOTYPES-GUIDE.md`
- `AGENTIC-DEVELOPMENT-GUIDE.md`
- `FREYA-WORKFLOW-INSTRUCTIONS.md`
- `page-template.html`
- `PROTOTYPE-ANALYSIS.md`
@ -228,7 +228,7 @@ project-analysis/
| ❌ Generic | ✅ Specific |
|-----------|------------|
| `README.md` | `INTERACTIVE-PROTOTYPES-GUIDE.md` |
| `README.md` | `AGENTIC-DEVELOPMENT-GUIDE.md` |
| `INSTRUCTIONS.md` | `FREYA-WORKFLOW-INSTRUCTIONS.md` |
| `GUIDE.md` | `DESIGN-SYSTEM-GUIDE.md` |
| `template.yaml` | `work-file-template.yaml` |

View File

@ -141,8 +141,8 @@ topic/
**Example**:
```
interactive-prototypes/
├── INTERACTIVE-PROTOTYPES-GUIDE.md (complete system overview)
agentic-development/
├── AGENTIC-DEVELOPMENT-GUIDE.md (complete system overview)
├── CREATION-GUIDE.md (detailed technical reference)
└── PROTOTYPE-ANALYSIS.md (case study)
```

View File

@ -0,0 +1,212 @@
# Agentic Development - File Index
**Location**: `src/workflows/4-ux-design/agentic-development/`
---
## 📁 Complete File Structure
```
agentic-development/
├── AGENTIC-DEVELOPMENT-GUIDE.md ← START HERE (overview & quick reference)
├── workflow.md ← Workflow overview with phase links
├── PROTOTYPE-INITIATION-DIALOG.md ← Conversation scripts for initiation
├── CREATION-GUIDE.md ← Original detailed guide (reference)
├── PROTOTYPE-ANALYSIS.md ← Dog Week analysis (examples)
├── steps-c/ ← Micro-step workflow files
│ ├── 1-prototype-setup.md ← Phase 1: Environment setup
│ ├── 2-scenario-analysis.md ← Phase 2: Analyze spec & create views
│ ├── 3-logical-view-breakdown.md ← Phase 3: Break view into sections
│ ├── 4a-announce-and-gather.md ← Phase 4a: Announce section
│ ├── 4b-create-story-file.md ← Phase 4b: Create story file
│ ├── 4c-implement-section.md ← Phase 4c: Implement code
│ ├── 4d-present-for-testing.md ← Phase 4d: Present for testing
│ ├── 4e-handle-issue.md ← Phase 4e: Fix issues (loop)
│ ├── 4f-handle-improvement.md ← Phase 4f: Handle improvements (loop)
│ ├── 4g-section-approved.md ← Phase 4g: Section approved
│ └── 5-finalization.md ← Phase 5: Integration test & approval
├── templates/
│ ├── work-file-template.yaml ← Planning document template
│ ├── story-file-template.md ← Section implementation template
│ ├── page-template.html ← Complete HTML page template
│ ├── PROTOTYPE-ROADMAP-template.md ← Scenario roadmap template
│ ├── demo-data-template.json ← Demo data structure template
│ └── components/
│ ├── dev-mode.html ← Dev mode toggle button
│ ├── dev-mode.js ← Dev mode logic (Shift+Click to copy IDs)
│ ├── dev-mode.css ← Dev mode styles
│ └── DEV-MODE-GUIDE.md ← Dev mode usage guide
└── examples/
└── (Dog Week prototypes as reference)
```
---
## 📚 What Each File Does
### Core Documentation
#### `AGENTIC-DEVELOPMENT-GUIDE.md`
**Purpose**: Complete system overview
**For**: All agents (Freya, Saga, Idunn)
**Contains**:
- System overview
- Folder structure
- Complete workflow summary
- Key principles
- Quick reference
- Success metrics
**Read this**: To understand the complete system
---
#### `workflow.md`
**Purpose**: Workflow overview with phase navigation
**For**: Freya (primary), other agents (reference)
**Contains**:
- Overview of all phases
- Clear links to step files
- When to use each phase
- What each phase creates
**Read this**: To understand the workflow structure
---
### Step Files
#### `steps-c/1-prototype-setup.md`
**Purpose**: Environment setup instructions
**Contains**: Device compatibility, design fidelity, languages, demo data creation
**Next**: Phase 2
---
#### `steps-c/2-scenario-analysis.md`
**Purpose**: Scenario analysis and view identification
**Contains**: Spec analysis, logical view mapping
**Next**: Phase 3
---
#### `steps-c/3-logical-view-breakdown.md`
**Purpose**: Break view into implementable sections
**Contains**: Section breakdown, work file creation
**Next**: Phase 4
---
#### `steps-c/4a-4g-*.md` (Phase 4 Loop)
**Purpose**: Section-by-section implementation
**Contains**: Announce, create story, implement, test, handle feedback, approve
**Flow**: 4a → 4b → 4c → 4d → [4e/4f loop] → 4g → [next section]
---
#### `steps-c/5-finalization.md`
**Purpose**: Integration test and completion
**Contains**: Final test, quality checklist, next steps
**Next**: New page (Phase 3) or new scenario (Phase 1)
---
### Templates
#### `templates/work-file-template.yaml`
**Purpose**: Planning document
**When to use**: Start of EVERY implementation
**Created**: Once per page at beginning
**Contains**:
- Metadata (page info, device compatibility)
- Design tokens (Tailwind config)
- Page requirements (from spec)
- Demo data needs
- Object ID map
- Section breakdown (4-8 sections)
- Testing checklist
**Use this**: To create work file (plan BEFORE coding)
---
#### `templates/story-file-template.md`
**Purpose**: Section implementation guide
**When to use**: Just-in-time (right before implementing each section)
**Created**: Once per section (4-8 per page)
**Contains**:
- Section goal
- What to build (HTML/JS)
- Tailwind classes to use
- Dependencies
- Acceptance criteria
- Test instructions
- Common issues
**Use this**: To create story file before each section
---
#### `templates/page-template.html`
**Purpose**: Complete HTML page structure
**When to use**: Creating new HTML page
**Created**: Once per page (at start of Section 1)
**Contains**:
- Complete HTML structure
- Tailwind CDN setup
- Tailwind config inline
- Component examples
- Shared script includes
**Use this**: As starting point for new page HTML
---
## 🎯 Which File When?
### Starting New Scenario
1. Read: `workflow.md` (understand phases)
2. Follow: `steps-c/1-prototype-setup.md` (setup)
3. Use: `PROTOTYPE-ROADMAP-template.md` → Create roadmap
4. Use: `demo-data-template.json` → Create demo data
### Starting New Page
1. Follow: `steps-c/2-scenario-analysis.md` (analyze)
2. Follow: `steps-c/3-logical-view-breakdown.md` (break down)
3. Use: `work-file-template.yaml` → Create work file
4. Get approval
### Implementing Each Section
1. Follow: `steps-c/4a-4g-*.md` (loop)
2. Use: `story-file-template.md` → Create story file (just-in-time)
3. Implement in HTML (incrementally)
4. Test
5. Get approval
6. Repeat for next section
### Finishing Page
1. Follow: `steps-c/5-finalization.md` (integration test)
2. Get final approval
3. Choose: New page, new scenario, or done
---
## 📝 Template Usage Summary
| Template | When Created | How Many | Purpose |
|----------|--------------|----------|---------|
| work-file | Start of page | 1 per page | Complete plan |
| story-file | Before each section | 4-8 per page | Section implementation |
| page | Start of Section 1 | 1 per page | HTML structure |
| roadmap | Start of scenario | 1 per scenario | Scenario overview |
| demo-data | Setup scenario | 1 per scenario | Auto-loading data |
---
**All templates and micro-step instructions are ready!**
Next step: Activate Freya and follow `workflow.md``steps-c/1-prototype-setup.md`

View File

@ -1,7 +1,7 @@
---
validationDate: 2026-01-22
workflowName: interactive-prototypes
workflowPath: c:\dev\WDS\bmad-method-wds-expansion\src\workflows\4-ux-design\interactive-prototypes
workflowName: agentic-development
workflowPath: c:\dev\WDS\bmad-method-wds-expansion\src\workflows\4-ux-design\agentic-development
validationStatus: COMPLETE
overallStatus: PASS - FIXED
criticalIssues: 2
@ -10,7 +10,7 @@ warnings: 1
warningsFixed: 1
---
# Validation Report: Interactive Prototypes
# Validation Report: Agentic Development
**Validation Started:** 2026-01-22
**Validator:** Wendy - BMAD Workflow Validation System
@ -30,7 +30,7 @@ warningsFixed: 1
### Current Structure
```
interactive-prototypes/
agentic-development/
├── phases/ ← ❌ Should be steps-c/
│ ├── 1-prototype-setup.md (95 lines) ✅
│ ├── 2-scenario-analysis.md (174 lines) ✅
@ -46,7 +46,7 @@ interactive-prototypes/
├── templates/ (template files) ✅
├── CREATION-GUIDE.md (1,148 lines)
├── FILE-INDEX.md (365 lines)
├── INTERACTIVE-PROTOTYPES-GUIDE.md (380 lines)
├── AGENTIC-DEVELOPMENT-GUIDE.md (380 lines)
├── PROTOTYPE-ANALYSIS.md (832 lines)
├── PROTOTYPE-INITIATION-DIALOG.md (409 lines)
└── PROTOTYPE-WORKFLOW.md (137 lines) ← ❌ Should be workflow.md
@ -87,7 +87,7 @@ interactive-prototypes/
**Current:** Guide files in root folder
- CREATION-GUIDE.md
- INTERACTIVE-PROTOTYPES-GUIDE.md
- AGENTIC-DEVELOPMENT-GUIDE.md
- PROTOTYPE-ANALYSIS.md
- PROTOTYPE-INITIATION-DIALOG.md
- FILE-INDEX.md
@ -147,7 +147,7 @@ interactive-prototypes/
**Action:** Move guide files to `guides/` subfolder
**Files to move:**
- CREATION-GUIDE.md
- INTERACTIVE-PROTOTYPES-GUIDE.md
- AGENTIC-DEVELOPMENT-GUIDE.md
- PROTOTYPE-ANALYSIS.md
- PROTOTYPE-INITIATION-DIALOG.md
- FILE-INDEX.md
@ -207,7 +207,7 @@ This is a **well-designed workflow** with **structural compliance issues**. The
3. ✅ **Organized guides into guides/ subfolder**
- Moved 5 guide files:
- CREATION-GUIDE.md
- INTERACTIVE-PROTOTYPES-GUIDE.md
- AGENTIC-DEVELOPMENT-GUIDE.md
- PROTOTYPE-ANALYSIS.md
- PROTOTYPE-INITIATION-DIALOG.md
- FILE-INDEX.md
@ -219,7 +219,7 @@ This is a **well-designed workflow** with **structural compliance issues**. The
### New Structure:
```
interactive-prototypes/
agentic-development/
├── workflow.md (entry point) ✅
├── steps-c/ (BMAD v6 compliant) ✅
│ ├── 1-prototype-setup.md
@ -236,7 +236,7 @@ interactive-prototypes/
├── guides/ (organized documentation) ✅
│ ├── CREATION-GUIDE.md
│ ├── FILE-INDEX.md
│ ├── INTERACTIVE-PROTOTYPES-GUIDE.md
│ ├── AGENTIC-DEVELOPMENT-GUIDE.md
│ ├── PROTOTYPE-ANALYSIS.md
│ └── PROTOTYPE-INITIATION-DIALOG.md
└── templates/ (existing templates) ✅

View File

@ -0,0 +1,631 @@
---
name: Agentic Development
description: AI-assisted development for non-technical designers using Agent Dialogs and structured collaboration
web_bundle: true
---
# Agentic Development Workflow
**Goal:** Enable non-technical designers to build production-ready code through structured AI collaboration
**Your Role:** Implementation partner guiding step-by-step development with clear feedback protocols and approval gates
---
## OVERVIEW
Agentic Development is about **how** you work with AI — not **what** you build.
Use this workflow to build a prototype, extend a production app, create a new feature, or anything else. The AI handles implementation while you focus on what to build and how it should behave.
This workflow enables:
- ✅ Direct implementation from specifications
- ✅ Step-by-step development with approval gates
- ✅ Clear feedback protocol with timing guidance
- ✅ Agent Dialog structure for context and handoff
- ✅ Dynamic planning that adapts as we learn
**Note:** We use "scenario step" instead of "page" - a step can be a full page, modal, overlay, state change, or any UI change requiring a new sketch.
---
## WHEN TO USE
Use this workflow when:
- ✅ Page specifications are complete and approved
- ✅ Ready to build working implementations
- ✅ Working with AI to develop production-ready code
- ✅ Want iterative development with approval gates
- ✅ Need structured collaboration with clear feedback handling
Skip this workflow when:
- ❌ Specifications not complete yet
- ❌ Still in sketching or wireframe phase
- ❌ Simple one-file changes that don't need documentation
- ❌ Pure exploration where the path is unclear
---
## USER FEEDBACK PROTOCOL
During development, the designer provides feedback. **CRITICAL: Never implement feedback without first classifying it and stating when it should be addressed.**
### Feedback Types
| Type | What It Is | When to Address |
|------|------------|-----------------|
| **Bug/Issue** | Something broken, error, not working | Now — fix immediately, iterate until resolved |
| **Quick Adjustment** | Small tweak, change X to Y | Now — implement immediately |
| **Addition** | New requirement that fits current dialog | Later step — add to plan |
| **Change Request** | Outside current dialog scope | Future session — document in Change Requests |
### The 2-Minute Rule (GTD)
**If a fix takes less than 2 minutes, do it immediately.**
From David Allen's "Getting Things Done": planning overhead should not exceed task complexity. See [GTD Model](../../../docs/models/gtd-getting-things-done.md).
| Situation | Action |
|-----------|--------|
| Missing condition check | Fix now, log as sub-step |
| Wrong variable name | Fix now, log as sub-step |
| Needs new component | Add to plan |
| Architectural change | Add to plan |
**Pattern:** Do the fix → Log as sub-step (e.g., 20a-1) → Continue main task
### Response Flow
**When user reports something:**
1. **CLASSIFY** — What type of feedback is this?
2. **TIMING** — When should it be addressed?
3. **DOCUMENT** — For bugs, add to plan BEFORE fixing
4. **CONFIRM** — For additions and change requests, confirm before proceeding
5. **EXECUTE** — Implement or document as appropriate
### Bug/Issue (Document First, Then Fix)
**User says:** "This is broken" / "Error occurred" / "Not working"
**Agent response:**
```
This is a bug — [brief description of issue].
Let's call it 10A and let me add it to the plan, then fix it.
```
**Required sequence:**
1. **Identify** — Understand and describe the bug clearly
2. **Document** — Add to dialog plan as sub-step (e.g., 21a, 21b)
3. **Execute** — Fix and iterate until resolved
4. **Log** — Record resolution in progress log
**If user reports multiple issues at once:**
1. Identify and list each issue separately
2. Add ALL to the plan first
3. Then work through them one by one
**Why document first:** Bugs discovered during implementation are valuable context. Recording them ensures traceability, prevents forgotten fixes, and creates a clear record for handoff.
**Example:**
```
User: "The drawer appears behind the overlay and the reassign doesn't work"
Agent: "I see two issues:
1. Drawer z-index (appears behind overlay)
2. Reassign not working
Let me add both to the plan as 21a and 21b, then fix them."
```
### Quick Adjustment (Fix Now)
**User says:** "Change X to Y" / "Make this button go here" / "Use different color"
**Agent response:**
```
Quick adjustment — I'll implement this now.
```
Then implement the change.
### Addition (Add to Plan)
**User says:** "We should also add X" / "Don't forget to handle Y"
**Agent response:**
```
This is an addition that fits the current dialog.
I'll add it to Step {N} for later. Confirm?
```
Then add to the appropriate step or To Do section in the dialog file.
### Change Request (Document for Later)
**User says:** "The profile button should go to /family" / "We need a settings page"
**Agent response:**
```
This is outside the current dialog scope.
It doesn't block {feature name}.
I'll add it to Change Requests for a future session. Confirm?
```
**WAIT for user confirmation.**
If user confirms → Add to Change Requests section
If user says "do it now" → Treat as quick adjustment and implement
### Anti-Pattern: What NOT to Do
**NEVER do this:**
User: "The profile button should go to /family"
Agent: *immediately edits the file*
**ALWAYS do this:**
User: "The profile button should go to /family"
Agent: "This is outside current scope. I'll add it to Change Requests. Confirm?"
User: "Yes" or "No, do it now"
*then acts accordingly*
### Why This Matters
From experience, skipping the classify-and-timing step leads to:
- Agent implementing things the user didn't want
- User frustration from lack of visibility
- Lost context when user wanted to discuss first
- Scope creep pulling focus from current implementation
The extra seconds to classify and confirm build trust and ensure alignment.
---
## SESSION START PROTOCOL
When starting or resuming a session, **always follow this sequence before implementing anything:**
### 1. Read the Dialog Document
```
Agent: "Let me read the dialog document to understand where we are."
```
Read the dialog file completely to understand:
- What steps are done
- What steps remain
- Any blockers or change requests
- Current context and decisions
### 2. Verify Plan Against Reality
**The plan may be outdated.** Check if:
- Steps marked "To Do" have actually been implemented
- Steps marked "Done" are truly complete
- The numbering is sequential and accurate
```
Agent: "I see steps 11-18 in To Do, but let me check the code..."
Agent: "These are already implemented. Let me update the plan."
```
If the plan is outdated → Update it before proceeding.
### 3. Present Current Status
Summarize for the designer:
- What's done (with step numbers)
- What's remaining (with step numbers)
- Any change requests pending
```
Agent: "Here's where we are:
Done: Steps 9b, 10, 11-19
To Do: Steps 20, 21
Ready to continue?"
```
### 4. Before Implementing a Step
**Always check the specification/sketches first:**
```
Agent: "Before implementing step 20, let me check the sketches..."
Agent: "I see this requires a nested drawer pattern, not inline buttons.
Should I break this into sub-steps?"
```
This prevents building the wrong thing and wasting effort.
### Why This Matters
Sessions can be interrupted. Context can be lost. The dialog document survives — but only if it's kept accurate. This protocol ensures:
- No duplicate work (re-implementing what exists)
- No missed work (skipping what's actually needed)
- Correct understanding of requirements before implementation
---
## EXECUTION PRINCIPLES
### Document Before Acting
**Every decision, action, and problem must be documented in the dialog file BEFORE acting on it.**
This ensures:
- Full traceability of what was decided and why
- Clean handoff if context is lost or session changes
- The dialog document is always the source of truth for progress
### Sketch Fidelity
**Implement code as close to the provided sketches as possible.**
Sketches are intentional design decisions, not loose suggestions. Text sizes, proportions, spacing, and layout choices are made for a reason:
| Element | Approach |
|---------|----------|
| **Text sizes** | Match relative sizes in sketch (headings vs body vs labels) |
| **Proportions** | Preserve ratios between elements |
| **Spacing** | Maintain visual rhythm and whitespace |
| **Layout** | Follow the arrangement precisely |
| **Component style** | Match the visual pattern (pills, cards, buttons) |
**When in doubt:**
1. Ask the designer for clarification
2. Reference the sketch directly in your question
3. Don't assume "close enough" is acceptable
**If constraints make exact matching impossible** (e.g., responsive behavior, platform limitations), document the deviation and explain why.
### Sub-Steps During Execution
While working on a step, you may discover additional tasks needed. Add these as sub-steps:
```markdown
| # | Section | Status | Notes |
|---|---------|--------|-------|
| 14 | Book It Button | ✅ | Complete |
| 14a | Fix button alignment | ✅ | Added during 14 |
| 14b | Add loading state | ✅ | Added during 14 |
| 15 | Cancel Button | 🔄 | In progress |
```
Sub-steps use letter suffixes (14a, 14b) to maintain the parent step's position.
### Dynamic Planning After Step Completion
**After completing each step, review and adjust the plan:**
1. **Review remaining steps** — Are they still accurate?
2. **Shuffle if needed** — Reorder based on what we learned
3. **Add new steps** — If implementation revealed new requirements
4. **Remove steps** — If no longer needed
5. **Update the dialog file** — Document the changes
**Numbering Rules:**
- **Completed steps:** Fixed numbering (never renumber)
- **Future steps:** Dynamic numbering (can change)
Example:
```markdown
### Done
| 9b | Carousel Refactor | ✅ | ← Fixed, never changes
| 10 | Deep Linking | ✅ | ← Fixed
### To Do
| 11 | Wire up handlers | 🔲 | ← Can be renumbered
| 12 | Add poop toggle | 🔲 | ← Can be removed or moved
| 13 | NEW: Error states | 🔲 | ← Can be inserted
```
### Handoff Always References Dialog
**Any handoff — to a new session, new agent, or human — MUST reference the dialog document.**
The dialog document is the single source of truth for:
- What has been done
- What decisions were made
- What remains to be done
- Any issues or blockers
Never hand off by describing the task verbally. Always point to the dialog.
### Plan-then-Execute Pattern
**Separate planning from execution into distinct sessions.**
Context windows are finite. Long sessions accumulate noise and risk context loss mid-implementation. The solution: end planning sessions deliberately, start execution sessions fresh.
**Planning Session:**
1. Explore the codebase and requirements
2. Discuss approach with the designer
3. Write the plan to the dialog file
4. End with a clear handoff: *"The plan is documented. Hasta la vista!"*
**Execution Session:**
1. Start fresh (new conversation)
2. Read the product brief for overall context
3. Read the page specification for requirements
4. Read the dialog document to understand the plan and progress
5. Execute the steps one by one
**Why This Works:**
- Fresh context window for execution = maximum working memory
- Dialog document carries all decisions and context forward
- No risk of context loss mid-implementation
- Each session has a clear, focused purpose
**When to Split:**
- After complex exploration or analysis
- When the plan is complete and approved
- When current session is getting long
- Before starting a major implementation phase
**The Dialog is the Bridge:**
```
┌─────────────────────┐ Dialog ┌─────────────────────┐
│ PLANNING SESSION │─────File────▶│ EXECUTION SESSION │
│ │ │ │
│ • Explore codebase │ │ • Fresh context │
│ • Discuss approach │ │ • Read brief + spec │
│ • Write plan │ │ • Read dialog │
│ • "Hasta la vista!" │ │ • Execute steps │
└─────────────────────┘ └─────────────────────┘
```
---
## WORKFLOW ARCHITECTURE
This uses **multi-phase architecture** with iterative loops:
### Phase Structure
**Sequential Phases (1-3, 5):** Setup → Analysis → Selection → Finalization
**Iterative Phase (4):** Section implementation loop with 7 micro-tasks
### Critical Rules
- 🎯 **ALWAYS** complete Phase 1 setup before starting
- 📊 **ALWAYS** analyze scenario before selecting views
- 🔁 **ALWAYS** use section-by-section approach (Phase 4 loop)
- ✅ **ALWAYS** get approval before moving to next section
- 📝 **ALWAYS** create story files just-in-time (not upfront)
---
## THE 5 PHASES
### Phase 1: Prototype Setup
**When:** Starting new scenario prototype (one-time per scenario)
**What:** Set up prototype environment and folder structure
**Creates:**
- Prototype folder with complete structure
- Demo data files
- Roadmap document
- All working folders (work/, stories/, shared/, components/, etc.)
**Go to:** [steps-c/1-prototype-setup.md](steps-c/1-prototype-setup.md)
---
### Phase 2: Scenario Analysis
**When:** Setup complete, ready to start building (one-time per scenario)
**What:** Analyze all scenario steps and identify logical views
**Creates:**
- Logical View Map (maps steps to views)
- View identification and relationships
**Go to:** [steps-c/2-scenario-analysis.md](steps-c/2-scenario-analysis.md)
---
### Phase 3: Logical View Selection & Breakdown
**When:** User selects which logical view to build (per view)
**What:** Identify all objects and break view into sections
**Creates:**
- Work file with section breakdown
- Implementation sequence
**Go to:** [steps-c/3-logical-view-breakdown.md](steps-c/3-logical-view-breakdown.md)
---
### Phase 4: Section Story & Implementation Loop
**When:** Ready to build sections (iterative per section)
**What:** For each section - prepare, create story, implement, test, handle feedback, approve
**The 7 Micro-Tasks:**
1. **[4a: Announce & Gather](steps-c/4a-announce-and-gather.md)**
- Announce section and gather requirements
2. **[4b: Create Story File](steps-c/4b-create-story-file.md)**
- Create focused story file for this section
3. **[4c: Implement Section](steps-c/4c-implement-section.md)**
- Implement code following story
4. **[4d: Present for Testing](steps-c/4d-present-for-testing.md)**
- Present to user with test instructions
5. **[4e: Handle Issue](steps-c/4e-handle-issue.md)**
- Fix issues if user reports problems (loop back to 4d)
6. **[4f: Handle Improvement](steps-c/4f-handle-improvement.md)**
- Implement improvements if user suggests (loop back to 4d)
7. **[4g: Section Approved](steps-c/4g-section-approved.md)**
- Finalize approval and move to next section (back to 4a)
**Flow:** 4a → 4b → 4c → 4d → [4e or 4f if needed, loops to 4d] → 4g → [back to 4a for next section]
**Creates (per section):**
- Story file (just-in-time)
- Incremental updates to view HTML
- Learnings captured
**Key:** One clear task per file → No confusion → Linear execution → Better results!
---
### Phase 5: Finalization
**When:** All sections complete for a logical view (end of view)
**What:** Integration test all states and final approval
**Result:** Production-ready logical view handling all its states
**Go to:** [steps-c/5-finalization.md](steps-c/5-finalization.md)
---
## INITIALIZATION
### Guide References
**Process Guides:**
- [AGENTIC-DEVELOPMENT-GUIDE.md](guides/AGENTIC-DEVELOPMENT-GUIDE.md) - Overview and methodology
- [CREATION-GUIDE.md](guides/CREATION-GUIDE.md) - Technical implementation details
- [PROTOTYPE-INITIATION-DIALOG.md](guides/PROTOTYPE-INITIATION-DIALOG.md) - Conversation scripts
- [PROTOTYPE-ANALYSIS.md](guides/PROTOTYPE-ANALYSIS.md) - Quality standards
- [FILE-INDEX.md](guides/FILE-INDEX.md) - Complete file reference
**Templates:**
- templates/work-file-template.yaml
- templates/story-file-template.md
- templates/page-template.html
- templates/components/dev-mode.*
### First Step Execution
Load, read and execute `steps-c/1-prototype-setup.md` to begin workflow.
---
## OUTPUT
**Per Scenario:**
```
[Scenario-Number]-[Scenario-Name]-Prototype/
├── [View].html files (in root, one per logical view)
├── shared/ (ONE COPY of shared code)
├── components/ (ONE COPY of reusable components)
├── pages/ (page-specific scripts if complex)
├── data/ (demo data JSON files)
├── stories/ (section development files - created just-in-time)
├── work/ (planning files)
└── PROTOTYPE-ROADMAP.md
```
**Result:** Self-contained, production-ready interactive prototypes with:
- Clean HTML using Tailwind CSS
- Vanilla JavaScript components
- Demo data auto-loading
- All states implemented and tested
---
## PROTOTYPE FOLDER STRUCTURE
```
[Scenario-Number]-[Scenario-Name]-Prototype/
├── [Page].html files ← Logical view HTML files (root)
├── shared/ ← ONE COPY of shared code
├── components/ ← ONE COPY of reusable components
├── pages/ ← Page-specific scripts (if complex)
├── data/ ← Demo data JSON files
├── stories/ ← Section development files (JIT)
├── work/ ← Planning files
│ ├── Logical-View-Map.md ← Maps steps to views
│ └── [View]-Work.yaml ← Section breakdowns per view
└── PROTOTYPE-ROADMAP.md ← Overall roadmap
```
---
## ITERATIVE WORKFLOW
**Phase 1-2:** One-time setup and analysis per scenario
**Phase 3:** Repeat for each logical view in scenario
**Phase 4:** Repeat for each section in current view
- Inner loop: Repeat 4d-4e-4f until approved
**Phase 5:** Repeat for each logical view (finalization)
**Pattern:**
```
Setup → Analysis → [View Selection → [Section Loop*] → Finalization]*
```
---
## EXAMPLES
**Typical Scenarios:**
1. **E-commerce Checkout:** 5 views (cart, shipping, payment, review, confirmation)
2. **SaaS Onboarding:** 4 views (signup, profile, preferences, dashboard)
3. **Booking System:** 6 views (search, select, details, confirm, payment, confirmation)
Each view breaks into 3-8 sections depending on complexity.
---
## QUALITY PRINCIPLES
**Section-by-Section Approval:** Never implement entire view at once - break into sections with approval gates
**Just-In-Time Stories:** Create story files only when needed (4b), not upfront
**Incremental Implementation:** Each section builds on previous approved sections
**Demo Data:** Use realistic demo data for testing and validation
**Self-Contained:** Each prototype folder is complete and portable
---
## TROUBLESHOOTING
**Issue:** User feedback requires rework
**Solution:** Use Phase 4e (issues) or 4f (improvements) to handle, then loop back to 4d
**Issue:** Section too complex
**Solution:** Break down further in Phase 3 before starting Phase 4
**Issue:** Logical view unclear
**Solution:** Revisit Phase 2 analysis to refine view mapping
---
## NOTES
This workflow enables **non-technical designers to build with AI** — whether that's:
- A quick prototype for user testing
- New features in an existing production app
- A complete application from scratch
The methodology stays the same:
- Step-by-step with approval gates
- Clear feedback protocol ensures alignment
- Agent Dialogs provide structure and traceability
- AI handles implementation, designer focuses on what and how
---
_Agentic Development - AI-assisted development for non-technical designers_

View File

@ -1,365 +0,0 @@
# Interactive Prototypes - File Index
**Location**: `src/modules/wds/workflows/4-ux-design/interactive-prototypes/`
---
## 📁 Complete File Structure
```
interactive-prototypes/
├── INTERACTIVE-PROTOTYPES-GUIDE.md ← START HERE (overview & quick reference)
├── PROTOTYPE-WORKFLOW.md ← Workflow overview with phase links
├── PROTOTYPE-INITIATION-DIALOG.md ← Conversation scripts for initiation
├── CREATION-GUIDE.md ← Original detailed guide (reference)
├── PROTOTYPE-ANALYSIS.md ← Dog Week analysis (examples)
├── phases/ ← Micro-step workflow files
│ ├── 1-prototype-setup.md ← Phase 1: Prototype environment setup
│ ├── 2-per-page-planning.md ← Phase 2: Analyze spec & create work file
│ ├── 3-section-implementation.md ← Phase 3: Build section-by-section
│ └── 4-finalization.md ← Phase 4: Integration test & approval
├── templates/
│ ├── work-file-template.yaml ← Planning document template
│ ├── story-file-template.md ← Section implementation template
│ ├── page-template.html ← Complete HTML page template
│ ├── PROTOTYPE-ROADMAP-template.md ← Scenario roadmap template
│ ├── demo-data-template.json ← Demo data structure template
│ └── components/
│ ├── dev-mode.html ← Dev mode toggle button
│ ├── dev-mode.js ← Dev mode logic (Shift+Click to copy IDs)
│ ├── dev-mode.css ← Dev mode styles
│ └── DEV-MODE-GUIDE.md ← Dev mode usage guide
└── examples/
└── (Dog Week prototypes as reference)
```
---
## 📚 What Each File Does
### Core Documentation
#### `INTERACTIVE-PROTOTYPES-GUIDE.md`
**Purpose**: Complete system overview
**For**: All agents (Freya, Saga, Idunn)
**Contains**:
- System overview
- Folder structure
- Complete workflow summary
- Key principles
- Quick reference
- Success metrics
**Read this**: To understand the complete system
---
#### `PROTOTYPE-WORKFLOW.md`
**Purpose**: Workflow overview with phase navigation
**For**: Freya (primary), other agents (reference)
**Contains**:
- Overview of all 4 phases
- Clear links to phase-specific files
- When to use each phase
- What each phase creates
**Read this**: To understand the workflow structure
---
#### `phases/1-prototype-setup.md`
**Purpose**: Prototype environment setup instructions
**Contains**: Device compatibility, design fidelity, languages, demo data creation
**Next**: Phase 2
---
#### `phases/2-per-page-planning.md`
**Purpose**: Page analysis and work file creation
**Contains**: Spec analysis, section breakdown, work file creation
**Next**: Phase 3
---
#### `phases/3-section-implementation.md`
**Purpose**: Section-by-section building
**Contains**: Story creation, implementation, testing, approval loop
**Next**: Phase 4 or repeat for next section
---
#### `phases/4-finalization.md`
**Purpose**: Integration test and completion
**Contains**: Final test, quality checklist, next steps
**Next**: New page (Phase 2) or new scenario (Phase 1)
---
#### `PROTOTYPE-INITIATION-DIALOG.md`
**Purpose**: Conversation scripts for initiation
**For**: Freya (exact scripts to follow)
**Contains**:
- Scenario initiation questions
- Per-page section breakdown prompts
- Example complete exchange
**Read this**: For exact conversation patterns
---
#### `CREATION-GUIDE.md`
**Purpose**: Original detailed guide
**For**: Deep dives, specific techniques
**Contains**:
- Detailed file structure explanations
- Step-by-step creation process
- Component patterns
- Testing strategies
- Common patterns library
**Read this**: For detailed technical reference
---
#### `PROTOTYPE-ANALYSIS.md`
**Purpose**: Dog Week case study
**For**: Learning from examples
**Contains**:
- Analysis of Dog Week prototypes
- What works well
- Patterns to follow
- Lessons learned
- Quality metrics
**Read this**: To see real-world examples
---
### Templates
#### `templates/work-file-template.yaml`
**Purpose**: Planning document
**When to use**: Start of EVERY prototype
**Created**: Once per page at beginning
**Contains**:
- Metadata (page info, device compatibility)
- Design tokens (Tailwind config)
- Page requirements (from spec)
- Demo data needs
- Object ID map
- Section breakdown (4-8 sections)
- Testing checklist
**Use this**: To create work file (plan BEFORE coding)
---
#### `templates/story-file-template.md`
**Purpose**: Section implementation guide
**When to use**: Just-in-time (right before implementing each section)
**Created**: Once per section (4-8 per page)
**Contains**:
- Section goal
- What to build (HTML/JS)
- Tailwind classes to use
- Dependencies
- Acceptance criteria
- Test instructions
- Common issues
**Use this**: To create story file before each section
---
#### `templates/page-template.html`
**Purpose**: Complete HTML page structure
**When to use**: Creating new HTML page
**Created**: Once per page (at start of Section 1)
**Contains**:
- Complete HTML structure
- Tailwind CDN setup
- Tailwind config inline
- Header example
- Form examples (input, textarea, split button)
- Submit button with loading state
- Toast notification
- Error banner
- Modal example (commented)
- Shared script includes
- Inline JavaScript template
**Use this**: As starting point for new page HTML
---
#### `templates/PROTOTYPE-ROADMAP-template.md`
**Purpose**: Scenario overview document
**When to use**: Start of scenario development
**Created**: Once per scenario
**Contains**:
- Scenario overview
- Device compatibility details
- Folder structure explanation
- Shared resources documentation
- Component documentation
- Prototype status table
- Testing requirements
- Troubleshooting guide
**Use this**: To create roadmap for scenario
---
#### `templates/demo-data-template.json`
**Purpose**: Demo data structure
**When to use**: Setting up scenario demo data
**Created**: Once per scenario (modify as needed)
**Contains**:
- User object
- Family object
- Members array
- Dogs array (or other entities)
- All fields with examples
**Use this**: To create demo-data.json file
---
## 🎯 Which File When?
### Starting New Scenario
1. Read: `PROTOTYPE-WORKFLOW.md` (understand phases)
2. Follow: `phases/1-prototype-setup.md` (setup)
3. Use: `PROTOTYPE-ROADMAP-template.md` → Create roadmap
4. Use: `demo-data-template.json` → Create demo data
### Starting New Page
1. Follow: `phases/2-per-page-planning.md` (analyze)
2. Use: `work-file-template.yaml` → Create work file
3. Get approval
4. Follow: `phases/3-section-implementation.md`
### Implementing Each Section
1. Follow: `phases/3-section-implementation.md` (loop)
2. Use: `story-file-template.md` → Create story file (just-in-time)
3. Implement in HTML (incrementally)
4. Test
5. Get approval
6. Repeat for next section
### Finishing Page
1. Follow: `phases/4-finalization.md` (integration test)
2. Get final approval
3. Choose: New page, new scenario, or done
### Need Help
1. Check: `PROTOTYPE-WORKFLOW.md` (phase overview)
2. Check: `phases/[N]-*.md` (specific phase)
3. Check: `CREATION-GUIDE.md` (detailed reference)
4. Check: `PROTOTYPE-ANALYSIS.md` (examples)
---
## 📊 File Relationships
```
PROTOTYPE-WORKFLOW.md (overview)
├─ phases/1-prototype-setup.md
├─ phases/2-per-page-planning.md
├─ phases/3-section-implementation.md
└─ phases/4-finalization.md
PROTOTYPE-INITIATION-DIALOG.md
└─ Referenced by: phases/1-prototype-setup.md (conversation scripts)
work-file-template.yaml
└─ Used in: phases/2-per-page-planning.md
└─ Each section becomes: story-file-template.md (later)
story-file-template.md
└─ Used in: phases/3-section-implementation.md (just-in-time)
└─ Guides: Implementation in HTML
page-template.html
└─ Used in: phases/3-section-implementation.md (Section 1 only)
└─ Modified: Section by section
PROTOTYPE-ROADMAP-template.md
└─ Used in: phases/1-scenario-init.md
└─ Updated: As prototypes complete
```
---
## 🚀 Quick Start Paths
### Path 1: I want to understand the system
1. `INTERACTIVE-PROTOTYPES-GUIDE.md` (overview)
2. `PROTOTYPE-WORKFLOW.md` (workflow phases)
3. `PROTOTYPE-ANALYSIS.md` (examples)
### Path 2: I want to create my first prototype
1. `PROTOTYPE-WORKFLOW.md` (start here)
2. `phases/1-prototype-setup.md` (follow step-by-step)
3. `phases/2-per-page-planning.md` (next)
4. `phases/3-section-implementation.md` (build loop)
5. `phases/4-finalization.md` (finish)
### Path 3: I need specific technical details
1. `CREATION-GUIDE.md` (detailed techniques)
2. `PROTOTYPE-ANALYSIS.md` (real examples)
3. `page-template.html` (code examples)
### Path 4: I'm stuck on something
1. `phases/[current-phase].md` (specific phase help)
2. `CREATION-GUIDE.md` → Common Pitfalls section
3. `templates/components/DEV-MODE-GUIDE.md` (if dev mode issue)
---
## 📝 Template Usage Summary
| Template | When Created | How Many | Purpose |
|----------|--------------|----------|---------|
| work-file | Start of page | 1 per page | Complete plan |
| story-file | Before each section | 4-8 per page | Section implementation |
| page | Start of Section 1 | 1 per page | HTML structure |
| roadmap | Start of scenario | 1 per scenario | Scenario overview |
| demo-data | Setup scenario | 1 per scenario | Auto-loading data |
---
## ✅ Checklist: Do I Have Everything?
**For Freya to create prototypes**:
- [x] `INTERACTIVE-PROTOTYPES-GUIDE.md` (overview)
- [x] `PROTOTYPE-WORKFLOW.md` (workflow overview)
- [x] `phases/1-prototype-setup.md` (Phase 1)
- [x] `phases/2-per-page-planning.md` (Phase 2)
- [x] `phases/3-section-implementation.md` (Phase 3)
- [x] `phases/4-finalization.md` (Phase 4)
- [x] `PROTOTYPE-INITIATION-DIALOG.md` (conversation scripts)
- [x] `work-file-template.yaml`
- [x] `story-file-template.md`
- [x] `page-template.html`
- [x] `PROTOTYPE-ROADMAP-template.md`
- [x] `demo-data-template.json`
- [x] `templates/components/dev-mode.*` (dev mode feature)
**For learning**:
- [x] `CREATION-GUIDE.md` (detailed)
- [x] `PROTOTYPE-ANALYSIS.md` (examples)
**For reference**:
- [x] Dog Week examples (real prototypes)
---
**All templates and micro-step instructions are ready!** 🎉
Next step: Activate Freya and follow `PROTOTYPE-WORKFLOW.md``phases/1-prototype-setup.md`

View File

@ -1,290 +0,0 @@
---
name: Interactive Prototypes
description: Create production-ready interactive prototypes using phased section-by-section implementation
web_bundle: true
---
# Interactive Prototypes Workflow
**Goal:** Build self-contained, production-ready interactive prototypes from UX specifications
**Your Role:** Implementation facilitator guiding section-by-section prototype development with approval gates
---
## OVERVIEW
This workflow creates interactive prototypes using:
- ✅ Tailwind CSS (no separate CSS files)
- ✅ Vanilla JavaScript (components in shared folders)
- ✅ Section-by-section implementation (approval gates)
- ✅ Just-in-time story files (created as needed)
- ✅ Demo data auto-loading
**Note:** We use "scenario step" instead of "page" - a step can be a full page, modal, overlay, state change, or any UI change requiring a new sketch.
---
## WHEN TO USE
Use this workflow when:
- ✅ Page specifications are complete and approved
- ✅ Ready to build working interactive prototypes
- ✅ Need production-ready HTML/CSS/JS implementations
- ✅ Want iterative development with approval gates
Skip this workflow when:
- ❌ Specifications not complete yet
- ❌ Still in sketching or wireframe phase
- ❌ Using different prototyping tools (Figma, Adobe XD)
- ❌ Building final production code (this is prototype stage)
---
## WORKFLOW ARCHITECTURE
This uses **multi-phase architecture** with iterative loops:
### Phase Structure
**Sequential Phases (1-3, 5):** Setup → Analysis → Selection → Finalization
**Iterative Phase (4):** Section implementation loop with 7 micro-tasks
### Critical Rules
- 🎯 **ALWAYS** complete Phase 1 setup before starting
- 📊 **ALWAYS** analyze scenario before selecting views
- 🔁 **ALWAYS** use section-by-section approach (Phase 4 loop)
- ✅ **ALWAYS** get approval before moving to next section
- 📝 **ALWAYS** create story files just-in-time (not upfront)
---
## THE 5 PHASES
### Phase 1: Prototype Setup
**When:** Starting new scenario prototype (one-time per scenario)
**What:** Set up prototype environment and folder structure
**Creates:**
- Prototype folder with complete structure
- Demo data files
- Roadmap document
- All working folders (work/, stories/, shared/, components/, etc.)
**Go to:** [steps-c/1-prototype-setup.md](steps-c/1-prototype-setup.md)
---
### Phase 2: Scenario Analysis
**When:** Setup complete, ready to start building (one-time per scenario)
**What:** Analyze all scenario steps and identify logical views
**Creates:**
- Logical View Map (maps steps to views)
- View identification and relationships
**Go to:** [steps-c/2-scenario-analysis.md](steps-c/2-scenario-analysis.md)
---
### Phase 3: Logical View Selection & Breakdown
**When:** User selects which logical view to build (per view)
**What:** Identify all objects and break view into sections
**Creates:**
- Work file with section breakdown
- Implementation sequence
**Go to:** [steps-c/3-logical-view-breakdown.md](steps-c/3-logical-view-breakdown.md)
---
### Phase 4: Section Story & Implementation Loop
**When:** Ready to build sections (iterative per section)
**What:** For each section - prepare, create story, implement, test, handle feedback, approve
**The 7 Micro-Tasks:**
1. **[4a: Announce & Gather](steps-c/4a-announce-and-gather.md)**
- Announce section and gather requirements
2. **[4b: Create Story File](steps-c/4b-create-story-file.md)**
- Create focused story file for this section
3. **[4c: Implement Section](steps-c/4c-implement-section.md)**
- Implement code following story
4. **[4d: Present for Testing](steps-c/4d-present-for-testing.md)**
- Present to user with test instructions
5. **[4e: Handle Issue](steps-c/4e-handle-issue.md)**
- Fix issues if user reports problems (loop back to 4d)
6. **[4f: Handle Improvement](steps-c/4f-handle-improvement.md)**
- Implement improvements if user suggests (loop back to 4d)
7. **[4g: Section Approved](steps-c/4g-section-approved.md)**
- Finalize approval and move to next section (back to 4a)
**Flow:** 4a → 4b → 4c → 4d → [4e or 4f if needed, loops to 4d] → 4g → [back to 4a for next section]
**Creates (per section):**
- Story file (just-in-time)
- Incremental updates to view HTML
- Learnings captured
**Key:** One clear task per file → No confusion → Linear execution → Better results!
---
### Phase 5: Finalization
**When:** All sections complete for a logical view (end of view)
**What:** Integration test all states and final approval
**Result:** Production-ready logical view handling all its states
**Go to:** [steps-c/5-finalization.md](steps-c/5-finalization.md)
---
## INITIALIZATION
### Guide References
**Process Guides:**
- [INTERACTIVE-PROTOTYPES-GUIDE.md](guides/INTERACTIVE-PROTOTYPES-GUIDE.md) - Overview and methodology
- [CREATION-GUIDE.md](guides/CREATION-GUIDE.md) - Technical implementation details
- [PROTOTYPE-INITIATION-DIALOG.md](guides/PROTOTYPE-INITIATION-DIALOG.md) - Conversation scripts
- [PROTOTYPE-ANALYSIS.md](guides/PROTOTYPE-ANALYSIS.md) - Quality standards
- [FILE-INDEX.md](guides/FILE-INDEX.md) - Complete file reference
**Templates:**
- templates/work-file-template.yaml
- templates/story-file-template.md
- templates/page-template.html
- templates/components/dev-mode.*
### First Step Execution
Load, read and execute `steps-c/1-prototype-setup.md` to begin workflow.
---
## OUTPUT
**Per Scenario:**
```
[Scenario-Number]-[Scenario-Name]-Prototype/
├── [View].html files (in root, one per logical view)
├── shared/ (ONE COPY of shared code)
├── components/ (ONE COPY of reusable components)
├── pages/ (page-specific scripts if complex)
├── data/ (demo data JSON files)
├── stories/ (section development files - created just-in-time)
├── work/ (planning files)
└── PROTOTYPE-ROADMAP.md
```
**Result:** Self-contained, production-ready interactive prototypes with:
- Clean HTML using Tailwind CSS
- Vanilla JavaScript components
- Demo data auto-loading
- All states implemented and tested
---
## PROTOTYPE FOLDER STRUCTURE
```
[Scenario-Number]-[Scenario-Name]-Prototype/
├── [Page].html files ← Logical view HTML files (root)
├── shared/ ← ONE COPY of shared code
├── components/ ← ONE COPY of reusable components
├── pages/ ← Page-specific scripts (if complex)
├── data/ ← Demo data JSON files
├── stories/ ← Section development files (JIT)
├── work/ ← Planning files
│ ├── Logical-View-Map.md ← Maps steps to views
│ └── [View]-Work.yaml ← Section breakdowns per view
└── PROTOTYPE-ROADMAP.md ← Overall roadmap
```
---
## ITERATIVE WORKFLOW
**Phase 1-2:** One-time setup and analysis per scenario
**Phase 3:** Repeat for each logical view in scenario
**Phase 4:** Repeat for each section in current view
- Inner loop: Repeat 4d-4e-4f until approved
**Phase 5:** Repeat for each logical view (finalization)
**Pattern:**
```
Setup → Analysis → [View Selection → [Section Loop*] → Finalization]*
```
---
## EXAMPLES
**Typical Scenarios:**
1. **E-commerce Checkout:** 5 views (cart, shipping, payment, review, confirmation)
2. **SaaS Onboarding:** 4 views (signup, profile, preferences, dashboard)
3. **Booking System:** 6 views (search, select, details, confirm, payment, confirmation)
Each view breaks into 3-8 sections depending on complexity.
---
## QUALITY PRINCIPLES
**Section-by-Section Approval:** Never implement entire view at once - break into sections with approval gates
**Just-In-Time Stories:** Create story files only when needed (4b), not upfront
**Incremental Implementation:** Each section builds on previous approved sections
**Demo Data:** Use realistic demo data for testing and validation
**Self-Contained:** Each prototype folder is complete and portable
---
## TROUBLESHOOTING
**Issue:** User feedback requires rework
**Solution:** Use Phase 4e (issues) or 4f (improvements) to handle, then loop back to 4d
**Issue:** Section too complex
**Solution:** Break down further in Phase 3 before starting Phase 4
**Issue:** Logical view unclear
**Solution:** Revisit Phase 2 analysis to refine view mapping
---
## NOTES
This workflow creates **prototypes**, not final production code. The code is production-ready quality but intended for:
- Stakeholder demos
- User testing
- Design validation
- Developer handoff reference
Final production code will be built in development phase using proper frameworks and architecture.
---
_Interactive Prototypes - Section-by-section implementation with approval gates_

View File

@ -572,7 +572,7 @@ This quality workflow integrates with:
- [Sketch Analysis](../substeps/4b-sketch-analysis.md) - Identifies page elements
**After:**
- [Interactive Prototypes](../interactive-prototypes/) - Builds HTML demos from specs
- [Agentic Development](../agentic-development/) - Builds HTML demos from specs
- [Design Deliveries](../../../6-design-deliveries/) - Packages specs for handoff
- [PRD Generation](../../../3-prd-platform/) - Creates developer stories from specs

View File

@ -0,0 +1,174 @@
# Stitch Prompt Template
Use this template to prepare an effective Stitch prompt from a WDS specification.
---
## How to Use
1. **Copy this template** into your Stitch dialog
2. **Fill in each section** using your spec and design system
3. **Remove Object IDs, translations, technical details** - Stitch doesn't need them
4. **Keep one language only** - typically the primary language (English or Swedish)
5. **Paste the filled template** as your Stitch prompt
---
## Template Structure
```
=== PROJECT CONTEXT ===
App: {App name} - {One-line description}
Target: {Target audience}
Brand feel: {2-3 adjectives describing the feel}
Market: {Market focus if relevant}
=== DESIGN SYSTEM ===
Colors:
- Background: {color name} ({hex})
- Primary/CTA: {color name} ({hex})
- Text: {color name} ({hex})
- Secondary text: {color name} ({hex})
- Success: {hex}
- Error: {hex}
Typography:
- Font: {font family}
- Headlines: {weight}, {characteristics}
- Body: {weight}, {size}
Component styles:
- Buttons: {style description - rounded, gradient, shadow, etc.}
- Inputs: {style description - border, focus state, etc.}
- Cards: {style description if relevant}
=== SCREEN DETAILS ===
Screen: {Screen name}
Purpose: {What this screen does, one sentence}
User context: {Where user is coming from, what they need}
Layout structure:
1. {Section 1}: {elements}
2. {Section 2}: {elements}
3. {Section 3}: {elements}
Key elements:
- {Element 1}: "{Actual content/text}"
- {Element 2}: "{Actual content/text}"
- {Element 3}: "{Actual content/text}"
Key interactions:
- Primary action: {what happens}
- Secondary action: {what happens}
=== CURRENT STATE NOTES ===
{Note any elements currently using default/unstyled components}
- {Component}: Currently ShadCN default, should match brand style
- {Component}: Uses custom gradient button
=== GENERATION INSTRUCTIONS ===
Generate this screen matching:
- Visual style of the attached reference image
- Layout structure of the attached sketch
- All content and elements listed above
Viewport: {Mobile 390px / Desktop 1440px}
```
---
## Example: Dog Week Sign-In
```
=== PROJECT CONTEXT ===
App: Dog Week - Family dog walk coordination app
Target: Swedish families (all ages from teens to grandparents)
Brand feel: Warm, friendly, trustworthy
Market: Sweden
=== DESIGN SYSTEM ===
Colors:
- Background: Cream (#FEF3CF), gradient to #FFFBED
- Primary/CTA: Orange (#FD6408), gradient #FD8002 to #FF2714
- Text: Brown (#2F1A0C)
- Secondary text: Gray (#686868)
- Success: Green (#28C54A)
- Error: Red (#DB0000)
Typography:
- Font: Inter
- Headlines: Bold/Extra Bold, tight letter spacing
- Body: Regular weight, 16px base
Component styles:
- Buttons: Rounded (8px), orange gradient for primary, subtle shadow
- Inputs: Light background, rounded corners, brown text
- Cards: Cream background, subtle shadow
=== SCREEN DETAILS ===
Screen: Sign In
Purpose: Authenticate users with email magic link or Google SSO
User context: Coming from Start Page, ready to access the app
Layout structure:
1. Header: Logo (left), Back button (right)
2. Main form: Email input, magic link button, divider, Google SSO
3. Trust section: Privacy and security messages
4. Help links: Support links at bottom
Key elements:
- Email input label: "Email address"
- Email placeholder: "your@email.com"
- Helper text: "We'll send you a magic link to sign in"
- Primary button: "Send magic link"
- Divider text: "Or sign in with"
- Google button: "Continue with Google"
- Trust message 1: "Your information is secure and private"
- Trust message 2: "We'll never spam you or share your details"
- Trust message 3: "Safe for all family members to use"
Key interactions:
- Primary: Enter email → Send magic link → Check email
- Secondary: Click Google → OAuth flow → Signed in
=== CURRENT STATE NOTES ===
- Input fields: Currently ShadCN default styling, should use cream background
- Google button: Should match brand's rounded style with Google colors
- Trust icons: Need checkmark or shield icons in success green
=== GENERATION INSTRUCTIONS ===
Generate this sign-in screen matching:
- Visual style of the attached Start Page screenshot (warm, cream, orange CTAs)
- Layout structure of the attached sketch
- All content and elements listed above
Viewport: Mobile 390px
```
---
## Checklist Before Pasting to Stitch
- [ ] Project context filled (app name, target, brand feel)
- [ ] Design system colors accurate (from Color-Palette.md)
- [ ] Typography correct (from Typography-System.md)
- [ ] Component styles described (buttons, inputs)
- [ ] Screen content in ONE language only (no translations)
- [ ] No Object IDs included
- [ ] No technical implementation details
- [ ] Current state notes added (what's ShadCN default)
- [ ] Viewport specified
---
_Stitch Prompt Template — Freya WDS Designer_

View File

@ -0,0 +1,288 @@
---
name: Stitch UI Generation
description: AI-assisted UI design using Google Stitch from specifications and sketches
web_bundle: true
---
# Stitch UI Generation
**Goal:** Generate production-quality UI designs using Google Stitch AI
**Your Role:** Guide the user through preparing inputs and creating a Stitch generation dialog
---
## OVERVIEW
Google Stitch is an AI-powered UI design tool that transforms text prompts, sketches, and reference images into responsive interfaces.
**Input Formula:**
```
Visual Reference + Sketch + Specification = Stitch Generation
```
**Output:** UI designs exportable to Figma or HTML/CSS
---
## WHEN TO USE STITCH
**Use Stitch when:**
- ✅ New page with detailed specification ready
- ✅ Have a visual reference (existing design or screenshot)
- ✅ Have a sketch showing layout structure
- ✅ Want rapid visual design iteration
- ✅ Need production-quality mockups quickly
**Skip Stitch when:**
- ❌ Building design system components (use tokens instead)
- ❌ Minor updates to existing designs (edit directly)
- ❌ No specification exists yet (write spec first)
- ❌ Complex multi-screen flows (do 2-3 screens at a time)
---
## BEFORE STARTING
### Prerequisites
1. **Specification exists** for the screen(s) to generate
2. **Visual reference available:**
- Screenshot of existing implemented page, OR
- Approved design that establishes visual language
3. **Sketch available** showing layout structure
### Decision: What Visual Reference?
| Situation | Reference Choice |
|-----------|------------------|
| Code exists and looks correct | Screenshot of running code |
| Starting fresh, no code | Original sketch or moodboard |
| Code exists but needs redesign | Original sketch |
| Want style from one page, layout from sketch | Both (upload 2 images) |
---
## WORKFLOW
### Step 1: Create Dialog
Create a Stitch generation dialog in the project's Agent Dialogs folder.
**Template:** Use `9-agent-dialogs/templates/dialog-types/stitch-generation.template.md`
**Location:** `docs/F-Agent-Dialogs/{YYYY-MM-DD}-freya-stitch-{feature}/`
**Dialog captures:**
- Which screens to generate
- What visual reference to use
- Pre-crafted prompts
- Generation results and decisions
---
### Step 2: Answer Pre-Generation Questions
For each potential screen, decide:
**Question 1: Generate in Stitch?**
| Screen | Has Code? | Has Sketch? | Generate? | Why |
|--------|-----------|-------------|-----------|-----|
| {screen} | ✅/❌ | ✅/❌ | ✅/❌ | {reason} |
**Question 2: What reference?**
| Screen | Reference | Source |
|--------|-----------|--------|
| {screen} | Code screenshot / Sketch | {path or action} |
---
### Step 3: Gather Inputs
For each screen to generate:
| Input | Action |
|-------|--------|
| **Visual Reference** | Take screenshot OR locate existing design |
| **Sketch** | Locate in spec's `Sketches/` folder |
| **Prompt** | Prepare using template (see Step 3a) |
---
### Step 3a: Prepare the Prompt
**DO NOT paste raw specifications into Stitch.** They contain:
- Object IDs (not useful for design generation)
- Multiple translations (confusing)
- Technical implementation details
**Instead, use the prompt template:** [stitch-prompt.template.md](stitch-prompt.template.md)
The prompt should include:
| Section | Source |
|---------|--------|
| **Project Context** | Design System overview, Product Brief |
| **Design System** | Color-Palette.md, Typography-System.md |
| **Component Styles** | Button/Input specs from Design System |
| **Screen Content** | Specification (ONE language, no Object IDs) |
| **Current State Notes** | What's ShadCN default vs custom |
**Important notes:**
- Use actual hex colors from your design system
- Describe component styles (rounded buttons, gradients, etc.)
- Include only ONE language (primary language)
- Note what's currently using default styling that needs brand styling
---
### Step 4: Generate in Stitch
1. Go to [stitch.withgoogle.com](https://stitch.withgoogle.com)
2. Upload **visual reference** image
3. Upload **sketch** image (if different from reference)
4. Paste **specification** as prompt
5. Add instruction: *"Generate this screen matching the visual style of the reference and layout of the sketch."*
6. Generate 2-3 variants
7. Select best result
**Settings:**
- Standard Mode: Quick iteration
- Pro Mode: Higher fidelity (uses more quota)
- Viewport: Mobile 390px or Desktop 1440px
---
### Step 5: Review Against Spec
| Check | Pass? |
|-------|-------|
| Content/copy matches spec | ✅/❌ |
| Layout follows sketch | ✅/❌ |
| Visual style matches reference | ✅/❌ |
| All key elements present | ✅/❌ |
**If issues:** Re-prompt with specific corrections or edit in Stitch.
---
### Step 6: Export & Store
| Format | When | Destination |
|--------|------|-------------|
| **Figma** | Team collaboration, design system work | Figma project |
| **HTML/CSS** | Code reference, rapid prototypes | `{spec-folder}/Visual-Design/` |
| **Screenshot** | Documentation | `{spec-folder}/Visual-Design/` |
**Naming:** `{screen-name}-stitch-v{#}.{ext}`
---
### Step 7: Update Specification
Add to the specification file:
```markdown
## Visual Design
**Stitch Generated:** [sign-in-stitch-v1.png](Visual-Design/sign-in-stitch-v1.png)
```
---
## STITCH CAPABILITIES & LIMITS
### What Stitch Does Well
- Single screen generation from prompts
- Style matching from reference images
- Responsive layouts (mobile/desktop)
- Clean HTML/CSS export
- Figma-compatible output
### Current Limitations
- Best with 2-3 screens at a time
- Layouts can be generic (need refinement)
- Doesn't capture deep UX nuance
- No built-in design system awareness
- Free tier: 350 standard + 200 pro screens/month
---
## PROMPT TIPS
### Effective Prompts Include
1. **App type:** "Mobile sign-in screen for..."
2. **Context:** "...a Swedish family dog walk coordination app"
3. **Visual direction:** "Match the attached reference..."
4. **Key elements:** List all major UI elements
5. **Content:** Include actual copy/text
6. **Mood:** "Friendly, trustworthy, family-oriented"
### Example Prompt Structure
```
[App type] [screen name] for [product description].
Visual style: Match the attached reference exactly.
Screen layout:
- [Section 1]: [elements and content]
- [Section 2]: [elements and content]
- [Section 3]: [elements and content]
Key interactions: [primary action], [secondary action]
Mood: [brand tone and feel]
Viewport: [Mobile 390px / Desktop 1440px]
```
---
## INTEGRATION WITH WDS
### Workflow Position
```
Specification → Stitch Generation → Visual Design → Implementation
↓ ↓ ↓ ↓
Content AI-Assisted Approved Code
& Structure Design Mockups Build
```
### Related Workflows
- **Before Stitch:** Page Specification workflow (create the spec first)
- **After Stitch:** Prototype Implementation dialog (build from the design)
- **Alternative:** Design Exploration dialog (if exploring multiple directions)
---
## QUICK START
**Fastest path:**
1. User: "I want to generate a UI design for [screen name]"
2. Freya: Creates Stitch dialog from template
3. Freya: Asks pre-generation questions (which screens, what reference)
4. User: Provides answers, takes screenshot if needed
5. User: Goes to Stitch, uploads inputs, pastes spec
6. User: Generates, reviews, exports
7. Freya: Updates dialog with results, links in spec
---
## SUPPORT
**Google Stitch:**
- [stitch.withgoogle.com](https://stitch.withgoogle.com)
- [Google Developers Blog - Introducing Stitch](https://developers.googleblog.com/stitch-a-new-way-to-design-uis/)
**WDS Templates:**
- Dialog template: `9-agent-dialogs/templates/dialog-types/stitch-generation.template.md`
---
_Stitch UI Generation Workflow — Freya WDS Designer — Phase 4 UX Design_

View File

@ -434,7 +434,7 @@ Area tags have limitations:
**Purpose:** Interactive tool for extracting Object IDs and area coordinates from prototypes
**Location:** `workflows/4-ux-design/interactive-prototypes/templates/components/dev-mode.js`
**Location:** `workflows/4-ux-design/agentic-development/templates/components/dev-mode.js`
### Features
@ -656,9 +656,9 @@ Area tags have limitations:
- W3C Spec: <https://www.w3.org/TR/html52/semantics-embedded-content.html#the-area-element>
**WDS Documentation:**
- Prototype Workflow: `workflows/4-ux-design/interactive-prototypes/`
- Prototype Workflow: `workflows/4-ux-design/agentic-development/`
- Figma Integration: `workflows/5-design-system/figma-integration/`
- Dev Mode: `workflows/4-ux-design/interactive-prototypes/templates/components/`
- Dev Mode: `workflows/4-ux-design/agentic-development/templates/components/`
---

View File

@ -2,7 +2,58 @@
## CONTEXT
This is the first step of the Agent Dialog Workflow. We create the folder structure and main dialog file.
This is the first step of the Agent Dialog Workflow. Before creating a new dialog, we first check for any existing pending dialogs.
---
## CHECK FOR EXISTING DIALOGS
⚠️ **IMPORTANT: Always check for pending dialogs before creating a new one.**
<action>
**Scan for existing dialogs:**
1. Check if `docs/F-Agent-Dialogs/` exists
2. If it exists, scan all subdirectories for dialog files (*-dialog.md)
3. Read the Meta section of each dialog file
4. Identify dialogs where:
- Status = "Not Started" OR "In Progress"
- Agent matches the current agent (if known)
</action>
<conditional>
**If pending dialogs found:**
Present them to the user:
```
📋 **Found {N} pending dialog(s):**
| # | Date | Agent | Feature | Status |
|---|------|-------|---------|--------|
| 1 | {date} | {agent} | {feature} | {status} |
**Options:**
[A] Continue existing dialog #{N}
[B] Create a new dialog (existing ones remain pending)
Choice:
```
If user chooses an existing dialog:
- Load that dialog file
- Check the Steps Overview table for next incomplete step
- Continue from there
If user chooses to create new:
- Continue to SELECT DIALOG TYPE below
</conditional>
<conditional>
**If no pending dialogs found OR docs/F-Agent-Dialogs/ doesn't exist:**
Continue to SELECT DIALOG TYPE below.
</conditional>
---
@ -70,12 +121,18 @@ Agent:</ask>
<action>
**Create dialog folder:**
Path: `docs/F-Agent-Dialogs/{date}-{feature_slug}/`
Path: `docs/F-Agent-Dialogs/{date}-{agent_slug}-{feature_slug}/`
Where:
- `{date}` = Today's date in YYYY-MM-DD format
- `{agent_slug}` = Agent name, lowercase (e.g., "freya", "saga", "dev")
- `{feature_slug}` = Feature name, lowercase, hyphenated
**Examples:**
- `2026-01-23-freya-booking-details-overlay/`
- `2026-01-23-saga-course-workflow-integration/`
- `2026-01-23-dev-calendar-scroll-fix/`
**Create subfolders:**
- `steps/` — For step instruction files
</action>
@ -87,9 +144,9 @@ Where:
<action>
**Create main dialog file:**
File: `docs/F-Agent-Dialogs/{date}-{feature_slug}/{date}-{feature_slug}-dialog.md`
File: `docs/F-Agent-Dialogs/{date}-{agent_slug}-{feature_slug}/{date}-{agent_slug}-{feature_slug}-dialog.md`
**Use template from:** `templates/dialog.template.md`
**Use template from:** Selected dialog type template (see SELECT DIALOG TYPE step above)
**Fill in:**
- Date: Today's date
@ -158,8 +215,8 @@ Patterns:</ask>
✅ **Dialog initialized!**
**Created:**
- Folder: `docs/F-Agent-Dialogs/{date}-{feature_slug}/`
- Dialog file: `{date}-{feature_slug}-dialog.md`
- Folder: `docs/F-Agent-Dialogs/{date}-{agent_slug}-{feature_slug}/`
- Dialog file: `{date}-{agent_slug}-{feature_slug}-dialog.md`
- Steps folder: `steps/`
**Next:** Proceed to Step 2 to analyze scope and create step breakdown.

View File

@ -4,6 +4,8 @@
Step files are created. Now we execute each step, either in this dialog or in fresh agent dialogs.
**Note:** For implementation-specific execution details (feedback protocol, dynamic planning), see the **Agentic Development Workflow** in `workflows/4-ux-design/agentic-development/workflow.md`.
---
## EXECUTION MODES
@ -34,9 +36,11 @@ Mode:</ask>
<action>
1. **Read the step file** completely
2. **Execute the implementation** following the instructions
3. **Verify against acceptance criteria**
4. **Test using the test instructions**
2. **Document any decisions or issues** in the dialog before implementing
3. **Execute the implementation** following the instructions
4. **Add sub-steps** if needed for things discovered during implementation
5. **Verify against acceptance criteria**
6. **Test using the test instructions**
</action>
### Step Completion Checklist
@ -73,7 +77,10 @@ Mode:</ask>
Choice:</ask>
<check if="choice == A">
<action>Load and execute next step file</action>
<action>
1. Review and update remaining steps in dialog file
2. Load and execute next step file
</action>
</check>
<check if="choice == B">
@ -81,6 +88,7 @@ Choice:</ask>
Update dialog file:
- Status: "In Progress"
- Progress Log: Note where we stopped
- Ensure all decisions documented
Output: "Progress saved. The dialog is ready to resume."
</action>
@ -90,6 +98,20 @@ Choice:</ask>
## MODE B: FRESH DIALOG INSTRUCTIONS
### Handoff Always References Dialog
<warning>
**Any handoff — to a new session, new agent, or human — MUST reference the dialog document.**
The dialog document is the single source of truth for:
- What has been done
- What decisions were made
- What remains to be done
- Any issues or blockers
Never hand off by describing the task verbally. Always point to the dialog.
</warning>
### For Each Step
<action>
@ -98,20 +120,24 @@ Choice:</ask>
```
## Execute Step {N}: {Step Name}
### Context
This is part of the {feature-name} implementation.
Dialog: docs/F-Agent-Dialogs/{date}-{feature-name}/
### Dialog Document (READ FIRST)
`docs/F-Agent-Dialogs/{date}-{feature-name}/{date}-{feature-name}-dialog.md`
Read the dialog file first to understand:
- Overall context and scope
- What steps are complete
- Current status and any blockers
### Step File
`docs/F-Agent-Dialogs/{date}-{feature-name}/steps/{N}-{step-name}.md`
### Instructions
1. Read the step file: `steps/{N}-{step-name}.md`
2. Execute the implementation
3. Verify against acceptance criteria
4. When complete, update the dialog file:
- Change step status to ✅
- Add entry to Progress Log
### Step File Location
`docs/F-Agent-Dialogs/{date}-{feature-name}/steps/{N}-{step-name}.md`
1. Read the dialog file completely
2. Read the step file
3. Execute the implementation
4. Document decisions in the dialog BEFORE implementing
5. Verify against acceptance criteria
6. Update dialog file: step status → ✅, add Progress Log entry
```
</action>
@ -197,7 +223,7 @@ Add to Progress Log:
<check if="all_steps_complete">
<output>
🎉 **All steps complete!**
**All steps complete!**
Proceed to Step 5 to finalize the dialog.

View File

@ -4,11 +4,24 @@ Choose the template that matches your work type.
---
## Naming Convention
```
docs/F-Agent-Dialogs/
└── {YYYY-MM-DD}-{agent}-{feature-name}/
└── {YYYY-MM-DD}-{agent}-{feature-name}-dialog.md
```
**Example:** `2026-01-23-freya-booking-details-overlay-dialog.md`
---
## Available Types
| Type | Icon | Template | Use When |
|------|------|----------|----------|
| **Prototype Implementation** | 🔧 | [prototype-implementation.template.md](prototype-implementation.template.md) | Building UI from specifications |
| **Stitch UI Generation** | 🪡 | [stitch-generation.template.md](stitch-generation.template.md) | AI-assisted UI design with Google Stitch |
| **Bug Fix** | 🐛 | [bug-fix.template.md](bug-fix.template.md) | Fixing issues and defects |
| **Design Exploration** | 🎨 | [design-exploration.template.md](design-exploration.template.md) | Exploring visual/UX directions |
| **Capture** | 💾 | [../dialog-capture.template.md](../dialog-capture.template.md) | Saving ideas for later |
@ -32,6 +45,25 @@ Choose the template that matches your work type.
---
### 🪡 Stitch UI Generation
**Best for:**
- Generating UI designs from specifications using Google Stitch AI
- Rapid visual design iteration
- Creating production-quality screens from sketches
**Key sections:**
- Input formula: Visual Reference + Sketch + Specification
- Screen-by-screen generation tracking
- Export and integration workflow
**Input Formula:**
```
Visual Reference + Sketch + Specification (as prompt) = Stitch Generation
```
---
### 🐛 Bug Fix
**Best for:**
@ -92,7 +124,9 @@ Choose the template that matches your work type.
```
Start here:
├── Building from spec? → 🔧 Prototype Implementation
├── Building code from spec? → 🔧 Prototype Implementation
├── Generating UI designs with AI? → 🪡 Stitch UI Generation
├── Fixing something broken? → 🐛 Bug Fix

View File

@ -1,4 +1,14 @@
# {DATE} {Feature Name} — Prototype Implementation
# {YYYY-MM-DD} {Feature Name} — Prototype Implementation
<!--
FILE NAMING CONVENTION:
- Folder: {YYYY-MM-DD}-{agent}-{feature-name}/
- File: {YYYY-MM-DD}-{agent}-{feature-name}-dialog.md
Example:
- Folder: 2026-01-23-freya-booking-details-overlay/
- File: 2026-01-23-freya-booking-details-overlay-dialog.md
-->
## Meta
@ -6,7 +16,7 @@
|-------|-------|
| **Date** | {YYYY-MM-DD} |
| **Type** | 🔧 Prototype Implementation |
| **Agent** | Freya (UX Designer) |
| **Agent** | {Agent name} ({Role}) |
| **Feature** | {Feature name} |
| **Specification** | [{Spec name}]({path-to-spec}) |
| **Target** | {Where to implement: app path or prototype folder} |
@ -126,24 +136,99 @@ Each section is implemented and approved before moving to the next:
### Approval Criteria
- [ ] Matches specification visually
- [ ] All Object IDs present
- [ ] **Matches sketches precisely** — Text sizes, proportions, spacing, layout
- [ ] All Object IDs present as `data-object-id` attributes
- [ ] Translations work (SE/EN toggle)
- [ ] Accessibility requirements met
- [ ] No TypeScript errors
> **Sketch Fidelity:** Sketches are intentional design decisions. Text sizes, proportions, and spacing are chosen deliberately. Implement as close to the sketch as possible. If constraints prevent exact matching, document the deviation.
---
## Feedback Protocol
How designer feedback is handled during implementation:
| Type | What It Is | When to Address | Documentation |
|------|------------|-----------------|---------------|
| **Bug/Issue** | Something broken or not working | Now — iterate until fixed | Part of current step |
| **Quick Adjustment** | Small tweak to current work | Now — implement immediately | Note in progress log |
| **Addition** | New requirement in scope | Later step — add to plan | Note in step file |
| **Change Request** | Outside current scope | Future session — document | Add to Change Requests |
### The 2-Minute Rule (GTD)
**If a fix takes less than 2 minutes, do it immediately.**
Planning overhead should not exceed task complexity. See [GTD Model](../../../../docs/models/gtd-getting-things-done.md).
**Pattern:** Do the fix → Log as sub-step (e.g., 20a-1) → Continue main task
### Agent Response Pattern
**When user reports something:**
1. **CLASSIFY** — What type of feedback is this?
2. **TIMING** — When should it be addressed?
3. **CONFIRM** — For additions and change requests, confirm before proceeding
4. **EXECUTE** — Implement or document as appropriate
### Change Request Flow
```
Designer: "The profile button should go to /family"
Agent: "This is outside the current dialog scope.
It doesn't block {current feature}.
I'll add it to Change Requests for a future session. Confirm?"
Designer: "Yes" → Agent adds to Change Requests
OR "Do it now" → Agent treats as quick adjustment, implements
```
---
## Change Requests
Structural changes identified during implementation. Assessed for timing.
| # | Request | Raised | Assessment | Timing | Status |
|---|---------|--------|------------|--------|--------|
| _None yet_ | | | | | |
<!--
Example:
| CR-1 | **Infinite Calendar Scroll** — Calendar should scroll continuously | 2026-01-24 | Structural change to calendar. Significant scope. | Future session | 🔲 Pending |
-->
---
## Steps Overview
| # | Section | Status | Approved |
|---|---------|--------|----------|
| 1 | [{Section name}](steps/01-{section}.md) | 🔲 | 🔲 |
| 2 | [{Section name}](steps/02-{section}.md) | 🔲 | 🔲 |
| 3 | [{Section name}](steps/03-{section}.md) | 🔲 | 🔲 |
### Done
| # | Section | Notes |
|---|---------|-------|
| _None yet_ | | |
### In Progress
| # | Section | Status | Description |
|---|---------|--------|-------------|
| 1 | [{Section name}](steps/01-{section}.md) | 🔄 | {Brief description} |
| 1a | — {Sub-task} | 🔄 | {Sub-step added during execution} |
### To Do
| # | Section | Description |
|---|---------|-------------|
| 2 | [{Section name}](steps/02-{section}.md) | {Brief description} |
| 3 | [{Section name}](steps/03-{section}.md) | {Brief description} |
**Status:** 🔲 Not Started | 🔄 In Progress | ✅ Complete | ⏸️ Blocked
**Approved:** 🔲 Pending | ✅ Approved | 🔁 Needs Changes
**Sub-Step Numbering:**
- Main steps: `1`, `2`, `3`...
- Sub-steps during execution: `1a`, `1b`...
- Bug fixes (2-min rule): `1a-1`, `1a-2`...
---

View File

@ -0,0 +1,311 @@
# {YYYY-MM-DD} {Feature Name} — Stitch UI Generation
<!--
FILE NAMING CONVENTION:
- Folder: {YYYY-MM-DD}-freya-stitch-{feature-name}/
- File: {YYYY-MM-DD}-freya-stitch-{feature-name}-dialog.md
Example:
- Folder: 2026-01-26-freya-stitch-signin-flow/
- File: 2026-01-26-freya-stitch-signin-flow-dialog.md
-->
## Meta
| Field | Value |
|-------|-------|
| **Date** | {YYYY-MM-DD} |
| **Type** | 🪡 Stitch UI Generation |
| **Agent** | Freya (WDS Designer) |
| **Feature** | {Feature/flow name} |
| **Specifications** | [{Spec 1}]({path}), [{Spec 2}]({path}) |
| **Design Reference** | [{Reference design}]({path}) — visual style source |
| **Status** | Not Started |
---
## Purpose
Generate production-quality UI designs for {feature name} using Google Stitch AI, based on existing specifications and sketches.
---
## About This Dialog
This dialog guides the **AI-assisted UI generation workflow** using Google Stitch.
```
┌─────────────────────┐ ┌─────────────────────┐ ┌─────────────────────┐
│ INPUTS │ │ STITCH PROCESS │ │ OUTPUTS │
│ │ │ │ │ │
│ • Specifications │────────▶│ • Prompt crafting │────────▶│ • UI designs │
│ • Sketches │ │ • Generation │ │ • Figma exports │
│ • Reference designs │ │ • Iteration │ │ • HTML/CSS code │
│ • Strategic context │ │ • Refinement │ │ • Design decisions │
└─────────────────────┘ └─────────────────────┘ └─────────────────────┘
Source Materials AI Generation Project Assets
```
**Specifications remain the single source of truth.** Stitch generates visual interpretations that must align with spec requirements.
---
## When to Use Stitch
| Scenario | Use Stitch? | Why |
|----------|-------------|-----|
| New page with detailed spec | ✅ Yes | Accelerates visual design from requirements |
| Exploring visual directions | ✅ Yes | Quick iteration on look & feel |
| Complex multi-state UI | ⚠️ Partially | Generate base, refine states manually |
| Design system components | ❌ No | Build from tokens, not generated |
| Minor spec updates | ❌ No | Edit existing designs directly |
---
## Stitch Capabilities & Limits
### What Stitch Does Well
- Single screen generation from prompts
- Style matching from reference images
- Responsive layouts (mobile/desktop)
- Clean HTML/CSS export
- Figma-compatible output
### Current Limitations
- Best with 2-3 screens at a time
- Layouts can be generic (need refinement)
- Doesn't capture deep UX nuance
- No built-in design system awareness
- Requires iteration for production quality
---
## Setup Context
### Project Context
- **Project:** {Project name}
- **Brand:** {Brand characteristics — tone, colors, typography}
- **Tech Stack:** {e.g., React, Next.js, Tailwind CSS}
- **Design System:** [{Design System}]({path})
### Strategic Context
| Document | Path | Key Elements |
|----------|------|--------------|
| VTC/Trigger Map | [{path}]({path}) | {Key triggers for this feature} |
| User Personas | [{path}]({path}) | {Target personas} |
| Product Brief | [{path}]({path}) | {Relevant product goals} |
### Reference Materials
| Type | File | Purpose |
|------|------|---------|
| **Style Reference** | [{design}]({path}) | Visual language source for Stitch |
| **Sketch** | [{sketch}]({path}) | Layout and structure guide |
| **Specification** | [{spec}]({path}) | Requirements and content |
---
## Pre-Generation Questions
Before generating, decide for each screen:
### Question 1: Which screens need Stitch generation?
| Screen | Has Code? | Has Sketch? | Generate in Stitch? | Why |
|--------|-----------|-------------|---------------------|-----|
| {Screen 1} | ✅/❌ | ✅/❌ | ✅/❌ | {reason} |
| {Screen 2} | ✅/❌ | ✅/❌ | ✅/❌ | {reason} |
### Question 2: What visual reference for each screen?
| Option | When to Use |
|--------|-------------|
| **Screenshot of existing code** | Code exists and represents the correct visual direction |
| **Original sketch** | Starting fresh, or code doesn't match desired direction |
| **Both** | Use code as style reference + sketch for layout guidance |
| Screen | Reference Choice | Source |
|--------|------------------|--------|
| {Screen 1} | Code screenshot / Sketch / Both | {path or "take screenshot"} |
| {Screen 2} | Code screenshot / Sketch / Both | {path or "take screenshot"} |
---
## Screens to Generate
| # | Screen | Specification | Visual Reference | Layout Guide | Priority | Status |
|---|--------|---------------|------------------|--------------|----------|--------|
| 1 | {Screen name} | [{spec}]({path}) | {screenshot/sketch} | {sketch if different} | High | 🔲 |
| 2 | {Screen name} | [{spec}]({path}) | {screenshot/sketch} | {sketch if different} | High | 🔲 |
**Status:** 🔲 Not Started | 🔄 In Progress | ✅ Generated | 🔁 Needs Refinement | ✅✅ Approved
---
## Stitch Input Formula
```
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ VISUAL REFERENCE│ + │ SKETCH │ + │ SPECIFICATION │ = Stitch Generation
│ (style source) │ │ (layout/structure)│ │ (paste as prompt)│
└─────────────────┘ └─────────────────┘ └─────────────────┘
```
**That's it.** Upload the reference + sketch images, paste the spec as prompt text.
---
## Stitch Workflow Steps
### Step 1: Gather Inputs
| Input | Source | Format |
|-------|--------|--------|
| **Visual Reference** | Screenshot of existing design (e.g., start page) | Image (PNG/JPG) |
| **Sketch** | Wireframe sketch from spec folder | Image (PNG/JPG) |
| **Specification** | Full spec markdown or key sections | Text (paste into prompt) |
**Checklist:**
- [ ] Visual reference captured (shows brand style)
- [ ] Sketch image ready
- [ ] Spec content copied (full doc or key sections)
---
### Step 2: Generate in Stitch
1. Go to [stitch.withgoogle.com](https://stitch.withgoogle.com)
2. Upload **visual reference** image
3. Upload **sketch** image
4. Paste **specification** as prompt text (or key sections)
5. Add brief instruction: *"Generate this screen matching the visual style of the reference and layout of the sketch."*
6. Generate 2-3 variants
7. Select best result
**Settings:**
- Mode: Standard (fast) or Pro (higher fidelity)
- Viewport: {Mobile 390px | Desktop 1440px}
**Results:**
| Screen | Variants | Selected | Notes |
|--------|----------|----------|-------|
| {name} | {#} | v{#} | {observations} |
---
### Step 3: Review Against Spec
Quick check - does the output match spec requirements?
| Check | ✅/❌ | Fix Needed |
|-------|-------|------------|
| Content/copy matches spec | | |
| Layout follows sketch | | |
| Visual style matches reference | | |
| All key elements present | | |
**If refinement needed:** Re-prompt with specific corrections or edit in Stitch.
---
### Step 4: Export & Store
| Format | Destination |
|--------|-------------|
| **Figma** | Copy to Figma for team/design system work |
| **HTML/CSS** | `{spec-folder}/Visual-Design/` |
| **Screenshot** | `{spec-folder}/Visual-Design/` |
**Naming:** `{screen-name}-stitch-v{#}.{ext}`
---
### Step 5: Link in Spec
Add reference to the specification file:
```markdown
## Visual Design
**Stitch Generated:** [{screen-name}-stitch-v1.png](Visual-Design/{screen-name}-stitch-v1.png)
```
---
## Prompts
Store each screen's prompt in a separate file in the dialog folder.
**Naming:** Match the scenario numbering: `{scenario#}-{screen-name}-stitch-prompt.md`
| # | Screen | Prompt File | Status |
|---|--------|-------------|--------|
| 1 | {Screen 1} | [{scenario#}-{screen-name}-stitch-prompt.md]({scenario#}-{screen-name}-stitch-prompt.md) | 🔲 |
| 2 | {Screen 2} | [{scenario#}-{screen-name}-stitch-prompt.md]({scenario#}-{screen-name}-stitch-prompt.md) | 🔲 |
**Examples:** `1.2-sign-in-stitch-prompt.md`, `1.3-profile-setup-stitch-prompt.md`
Use the [stitch-prompt.template.md](../../stitch-generation/stitch-prompt.template.md) to create each prompt file.
---
## Outputs Log
| Screen | Version | Format | Path | Approved? |
|--------|---------|--------|------|-----------|
| {name} | v1 | Figma | {link} | 🔲 |
| {name} | v2 | HTML | {path} | ✅ |
---
## Design Decisions
| Decision | Context | Rationale |
|----------|---------|-----------|
| {what was decided} | {why it came up} | {reasoning} |
---
## Spec Deviations
Issues where Stitch output differs from specification:
| Screen | Spec Requirement | Stitch Output | Resolution |
|--------|------------------|---------------|------------|
| {screen} | {requirement} | {what Stitch did} | {how resolved} |
---
## Progress Log
### {YYYY-MM-DD}
- Created dialog for {feature}
- Identified {#} screens to generate
- Prepared reference materials
<!--
### {YYYY-MM-DD}
- Generated Screen 1: {name} - v1
- Refinements needed: {list}
- Re-generated with updated prompt
- Approved ✅
-->
---
## Learnings
<!--
Capture prompt patterns and techniques that worked well.
-->
| What Worked | Why | Reuse For |
|-------------|-----|-----------|
| {technique} | {reason} | {future use} |
---
_Stitch UI Generation Dialog — Freya WDS Designer — WDS Agent Dialog Workflow_

View File

@ -104,6 +104,33 @@ This dialog **bridges the gap** between the page specification and the developme
---
## Implementation Workflow
### Step-by-Step Process
1. **Build step** — Implement the code
2. **Test step** — Verify against spec
3. **User approval** — Confirm it's correct
4. **CW status check** — Verify context window has room for next step
5. **Next step** — Move forward
### Context Window (CW) Status Check
At the end of each completed step, perform a CW status check to ensure the AI session can handle the next step without losing important context.
| Check | Action |
|-------|--------|
| **Session length** | If conversation is very long, consider starting fresh for next step |
| **Context clarity** | Can the agent still reference earlier decisions accurately? |
| **Handoff summary** | If starting new session, document: completed steps, key decisions, current file states |
**Recommendation thresholds:**
- **Continue** — Session is manageable, context is clear
- **Consider fresh start** — Session is long but context is still intact
- **Fresh start recommended** — Context overflow risk, or major architectural shift in next step
---
## Steps Overview
| # | Step | Status | Notes |

View File

@ -94,4 +94,25 @@ For each Object ID:
---
## CW Status Check
At step completion, assess context window status:
| Indicator | Status |
|-----------|--------|
| Session length | {Short / Medium / Long} |
| Context clarity | {Clear / Degraded} |
| Recommendation | {Continue / Consider fresh start / Fresh start recommended} |
**If starting fresh session for next step, include handoff summary:**
```
Completed: Step {#} - {Step Name}
Key decisions: {list any important decisions made}
File states: {list modified files and their current state}
Next: Step {#+1} - {Next Step Name}
```
---
_Step {#} of {total} — {Feature Name} Implementation_

View File

@ -388,12 +388,14 @@ To begin, load and execute [steps/step-01-initialize-dialog.md](steps/step-01-in
Agent Dialogs can be created from:
- **Interactive Prototypes Workflow** — Each section becomes a step
- **UX Design Workflow** — Specification implementation becomes a dialog
- **Design System Workflow** — Component creation becomes a dialog
- **Testing Workflow** — Validation tasks become a dialog
The Agent Dialog structure provides a consistent way to document and execute implementation work from any source.
**Note:** This workflow IS the Agentic Development approach. Interactive Prototypes, Prototype Implementation, Bug Fixes, and Design Exploration are all output types of this workflow.
---
_Agent Dialog Workflow — Structured, reproducible implementation with isolated context_

View File

@ -141,8 +141,8 @@ warningsFixed: 6
---
### 10. interactive-prototypes ⚠️ EMPTY/INCOMPLETE
**Location:** `4-ux-design/interactive-prototypes`
### 10. agentic-development ⚠️ EMPTY/INCOMPLETE
**Location:** `4-ux-design/agentic-development`
**Structure:** Directory exists but appears empty or undeveloped
**Issue:** No files found
**Recommendation:** Remove or develop workflow
@ -269,7 +269,7 @@ Individual validation reports created:
### Immediate Actions Required
1. **Create workflow.md files** for remaining workflows with step structures
2. **Investigate empty workflows** (interactive-prototypes, modular-architecture, object-types) - remove or develop
2. **Investigate empty workflows** (agentic-development, modular-architecture, object-types) - remove or develop
3. **Clarify guide vs workflow** distinction for excalidraw-integration and scenario-init
4. **Complete vtc-workshop** workflow.md creation