Claude Code Workflow

Nate Ober

Jan 6, 2026

Updated Jan 6, 2026

67129b2

Prompt Content

# Product Development Workflow for Claude Code & Cursor

> **Load this file to transform Claude into your product development partner.**
>
> This workflow guides you from idea to interactive prototype with professional HTML deliverables at each step.

---

## Quick Start

1. **Copy this folder** to your project: `cp -r prompts/ your-project/prompts/`
2. **Start Claude Code or Cursor** in your project directory
3. **Load this file** and describe your product idea
4. **Choose your mode**: Full Approval (pause after each phase) or Streamlined (work through all phases)

---

## Workflow Overview

```
Discovery → Market Research → [AI Framing] → PRFAQ → PRD → Prototype
                               (optional)
```

| Phase | Output | Guide to Load |
|-------|--------|---------------|
| 1 | Market Research | `Market Research Agent.md` |
| 1b | AI Framing (ML products only) | `AI Framing Agent.md` |
| 2 | PRFAQ | `PRFAQ Guide.md` |
| 3 | PRD | `PRD Creation Guide.md` |
| 4 | Prototype | `Prototype Creation Guide.md` |

**All outputs:** Standalone HTML files saved to `./documents/`

---

## Step 1: Discovery

### Ask the user:
- What problem are you trying to solve?
- Who is your target audience/customer?
- What are your main business goals?
- What key features do you envision?
- How will users interact with this product?
- Is this for an existing company? (for brand research)

### Inform user about workflow mode:
After gathering initial information, tell the user:
> "I'll work through 4 phases: Market Research → PRFAQ → PRD → Prototype. By default, I'll pause after each phase for your feedback. If you'd prefer I work through everything continuously, just say 'switch to streamlined' at any time."

### Determine if AI/ML product:
**AI/ML indicators:** ML models, predictions, NLP, computer vision, recommendations, automated decisions, pattern recognition, generative AI.

- **If AI/ML:** Include Phase 1b (AI Framing) after Market Research
- **If standard:** Skip Phase 1b, go directly to PRFAQ

### Workflow Modes

**Full Approval Mode (default):**
- After completing each phase, **STOP and wait for user approval** before proceeding
- Present a summary of what was created and ask: "Ready to proceed to the next phase, or would you like changes?"
- If you have questions or need clarification, **ask and wait for a response** before continuing
- Do NOT proceed to the next phase until the user explicitly approves

**Streamlined Mode:**
- Work through all phases continuously without stopping
- User can interrupt at any time to provide feedback
- If interrupted, pause and address feedback before resuming

> **Users can switch modes at any time** by saying "switch to streamlined" or "switch to full approval"

---

## Step 2: Execute Phases

### Phase 1: Market Research

**Load:** `prompts/Market Research Agent.md`

> **Important:** Use web search to find real competitor data, pricing, market reports, and customer reviews. Don't make up data.

**Conduct web-based research:**
1. **Competitive Landscape** (3-5 competitors)
   - Product offerings and positioning
   - Pricing models with actual figures
   - Strengths and weaknesses
2. **Market Sizing**
   - TAM (Total Addressable Market) with sources
   - SAM (Serviceable Addressable Market)
   - SOM (Serviceable Obtainable Market)
3. **Customer Pain Points**
   - From reviews, forums, social media
   - Specific quotes and examples
   - Unmet needs in current solutions
4. **Pricing Intelligence**
   - Competitor pricing tiers
   - Industry benchmarks
5. **Brand Research** (if building for existing company)
   - Logo, colors, typography
   - Brand voice and positioning

**Save:**
- `documents/MarketResearch_[Product]_[YYYY-MM-DD].html`

**Checkpoint:**
- [ ] TAM/SAM/SOM with dollar figures and sources
- [ ] At least 3 competitors with real pricing
- [ ] Pain points are specific, not generic
- [ ] No placeholder text (TBD, TODO, [insert])
- [ ] File saved successfully

> **Full Approval Mode:** STOP here. Present summary and wait for user approval before proceeding.

---

### Phase 1b: AI Framing (AI/ML Products Only)

**Load:** `prompts/AI Framing Agent.md`

**Create:**
1. Business goals and ML problem framing
2. ML problem type (classification, regression, etc.)
3. Input/output schemas
4. Data strategy and requirements
5. Performance metrics and thresholds
6. Feasibility assessment

**Save:**
- `documents/AIFraming_[Product]_[YYYY-MM-DD].html`

**Checkpoint:**
- [ ] ML problem type clearly defined
- [ ] Success metrics with specific thresholds
- [ ] Data requirements documented
- [ ] Feasibility assessment complete
- [ ] File saved successfully

> **Full Approval Mode:** STOP here. Present summary and wait for user approval before proceeding.

---

### Phase 2: PRFAQ

**Load:** `prompts/PRFAQ Guide.md`

**Incorporate Market Research findings, then create:**
1. Work through 5 Working Backwards questions:
   - Who is the customer? (use research insights)
   - What is the customer problem? (use pain points from research)
   - What is the solution?
   - What is the customer experience?
   - How will we measure success?
2. Write Press Release (as if product launched)
3. Write FAQ (address skeptical questions)

**Save:**
- `documents/PRFAQ_[Product]_[YYYY-MM-DD].html`

**Checkpoint:**
- [ ] Incorporates market research findings
- [ ] Compelling headline (not generic)
- [ ] Customer problem is specific with data
- [ ] Solution addresses researched pain points
- [ ] FAQ addresses real concerns (not softballs)
- [ ] File saved successfully

> **Full Approval Mode:** STOP here. Present summary and wait for user approval before proceeding.

---

### Phase 3: PRD

**Load:** `prompts/PRD Creation Guide.md`

**Create from PRFAQ and Market Research:**
1. User personas with detailed profiles
2. Requirements in EARS syntax (When/The/Shall)
3. User stories with acceptance criteria
4. Success metrics and business model
5. Technical constraints (AWS/Anthropic stack)
6. MLP Testing Plan (mandatory)

**Save:**
- `documents/PRD_[Product]_[YYYY-MM-DD].html`
- `documents/DesignSystem_[Product]_[YYYY-MM-DD].html`

**Checkpoint:**
- [ ] Personas based on market research customer insights
- [ ] Requirements traceable to PRFAQ
- [ ] Competitive positioning informed by research
- [ ] Testing plan included
- [ ] Design system created (follows design standards)
- [ ] Tech stack uses AWS-native services
- [ ] Files saved successfully

> **Full Approval Mode:** STOP here. Present summary and wait for user approval before proceeding.

---

### Phase 4: Prototype

**Load:** `prompts/Prototype Creation Guide.md`

**Create from PRD (modular structure required):**
1. **Design System first** - shared CSS tokens and components
2. User flow mapping
3. Information architecture
4. **Individual screen HTML files** (NOT one monolithic file)
5. Clickable prototype with navigation
6. Form validation and interactions
7. Project Dashboard (navigation hub)

> **Critical: Connect All Screens Together**
>
> Every button, link, and navigation element must work:
> - Use `href="Screen_[Name]_[Product]_[Date].html"` for links between screens
> - Navigation menus should link to all main screens
> - "Back" buttons should return to the previous screen
> - Form submissions should navigate to success/confirmation screens
> - Dashboard cards should link to their detail screens
> - User flows must be completable end-to-end by clicking through
>
> **Test every link** before marking the prototype complete.

**Save:**
- `documents/DesignSystem_[Product]_[YYYY-MM-DD].html` (create FIRST)
- `documents/ScreenIndex_[Product]_[YYYY-MM-DD].html` (navigation hub)
- `documents/Screen_[Name]_[Product]_[YYYY-MM-DD].html` (one per screen)
- `documents/ClickablePrototype_[Product]_[YYYY-MM-DD].html`
- `documents/ProjectDashboard_[Product]_[YYYY-MM-DD].html`

**Checkpoint:**
- [ ] Design system created FIRST with shared CSS
- [ ] Modular structure (separate files per screen)
- [ ] All PRD screens implemented
- [ ] **All buttons and links navigate to correct screens**
- [ ] **User flows completable end-to-end**
- [ ] Forms have validation
- [ ] Responsive on mobile/tablet/desktop
- [ ] Realistic data (no Lorem ipsum)
- [ ] Follows design standards (no AI slop)
- [ ] Files saved successfully

---

## File Naming Convention

```
[DocumentType]_[ProductName]_[YYYY-MM-DD].html
```

**Examples:**
- `MarketResearch_SmartInventory_2025-01-06.html`
- `PRFAQ_SmartInventory_2025-01-06.html`
- `PRD_SmartInventory_2025-01-06.html`
- `Screen_Dashboard_SmartInventory_2025-01-06.html`

**Rules:**
- Replace spaces with underscores
- Use title case for product name
- **HTML only** (no markdown files)
- Date is creation date

---

## File Structure After Completion

```
documents/
├── ProjectDashboard_[Product]_[Date].html    (navigation hub)
├── MarketResearch_[Product]_[Date].html      (Phase 1)
├── AIFraming_[Product]_[Date].html           (Phase 1b - AI/ML only)
├── PRFAQ_[Product]_[Date].html               (Phase 2)
├── PRD_[Product]_[Date].html                 (Phase 3)
├── DesignSystem_[Product]_[Date].html        (shared CSS)
├── ScreenIndex_[Product]_[Date].html         (screen navigation)
├── Screen_Dashboard_[Product]_[Date].html
├── Screen_[Name]_[Product]_[Date].html       (one per screen)
└── ClickablePrototype_[Product]_[Date].html
```

---

## Quality Checklist

### After Each Phase:
- [ ] HTML file saved with correct naming
- [ ] Content builds on previous phase
- [ ] Uses consistent design system styling
- [ ] No placeholder text (TBD, TODO, Lorem ipsum)
- [ ] Professional formatting and typography

### Cross-Phase Consistency:
- [ ] Market research informs PRFAQ customer problem
- [ ] Personas consistent across PRFAQ → PRD → Prototype
- [ ] Competitive positioning consistent throughout
- [ ] Success metrics coherent across documents
- [ ] Technical constraints carried forward

### Prototype Specific:
- [ ] **Modular structure** (NOT single monolithic file)
- [ ] Design system created FIRST
- [ ] Every PRD screen included
- [ ] All workflows completable end-to-end
- [ ] Mobile responsive on all screens
- [ ] Form validation works
- [ ] Navigation between all screens works
- [ ] Data consistent across screens
- [ ] **No AI slop** (see Design Standards)

---

## Design Standards

### Anti-Patterns (NEVER USE):
- Generic fonts: Inter, Roboto, Arial, system-ui
- Purple-to-blue gradients on white
- Uniform card grids
- Bootstrap/Tailwind defaults without customization
- Excessive emojis

### Required:
- Distinctive typography matching product aesthetic
- 60-30-10 color rule (dominant/secondary/accent)
- Visual texture (gradients, shadows, depth)
- Bouncy animations for key moments (custom cubic-bezier)
- Modular file structure for prototypes

### Aesthetic Directions:
| Product Type | Direction | Key Traits |
|--------------|-----------|------------|
| Enterprise B2B | Luxury/Refined | Serif fonts, gold accents, subtle shadows |
| Developer Tools | Retro-Futuristic | Dark mode, neon glows, monospace |
| Consumer Apps | Playful | Rounded corners, bouncy animations, bright |
| Content Platforms | Editorial | Strong typography, dramatic whitespace |
| Dashboards | Industrial | Dense data, functional, efficient |
| Wellness/Health | Organic | Earth tones, soft curves, calm |

See `samples/DesignSystem_TeenFit.html` for implementation example.

---

## AWS-Native Architecture

As an AWS-provided toolkit, technical designs prefer AWS services for enterprise-grade scalability, security, and compliance:

- **Compute:** Lambda, ECS, EC2, App Runner
- **Database:** DynamoDB, Aurora, RDS
- **Generative AI:** Amazon Bedrock, Bedrock AgentCore, Amazon Q
- **Storage:** S3, EFS
- **API:** API Gateway, AppSync
- **Auth:** Cognito

Amazon Bedrock provides access to foundation models from Amazon (Nova) and third-party providers (Anthropic Claude, Meta Llama, Mistral, and more).

---

## Context Integration

When the user provides additional context (CSV files, company docs, team info):

**Use it to:**
- Inform market research with real company data
- Create realistic personas from real team members
- Use actual customer names and metrics
- Incorporate specific industry terminology
- Reflect actual business scale
- Make scenarios match real workflows

**Quality standard:** All examples should feel authentic to the provided context.

---

## Switching Modes & Handling Interruptions

**User can switch modes anytime:**
- "Switch to streamlined" → Continue through remaining phases without pausing
- "Switch to full approval" → Pause after each phase for feedback

**If the user interrupts during Streamlined mode:**
1. Pause immediately
2. Show current progress
3. Ask: "What would you like to review or change?"
4. Offer options:
   - Review current phase
   - Make changes
   - Switch to Full Approval mode
   - Continue after addressing feedback
5. Resume based on user preference

---

## Validation Requirements by Phase

### Market Research Validation:
- TAM/SAM/SOM with actual dollar figures
- Sources cited for market data
- At least 3 competitors with real pricing
- Pain points must be specific quotes/examples

### PRFAQ Validation:
- Headline is compelling and specific
- Customer problem grounded in research
- Solution is concrete, not vague
- FAQ addresses skeptical questions

### PRD Validation:
- Requirements in EARS syntax
- Acceptance criteria for each story
- Testing plan included
- Tech stack compliance

### Prototype Validation:
- Modular file structure (not monolithic)
- Design system created first
- All screens from PRD implemented
- Navigation works between all screens
- No AI slop aesthetics

---

## Sample Outputs

See `samples/` folder for example outputs:
- `samples/PRFAQ_TeenFit.html`
- `samples/PRD_TeenFit.html`
- `samples/DesignSystem_TeenFit.html`
- `samples/Screen_Dashboard_TeenFit.html`

Open in browser to see quality standards.

---

## Troubleshooting

### Files not saving?
- Ensure `documents/` folder exists (create if needed)
- Use correct naming convention
- Verify file creation before proceeding

### Market research feels thin?
- Use web search to find real competitor data
- Look for actual pricing pages
- Search for customer reviews and complaints
- Find industry reports for market sizing

### Content feels generic?
- Review all provided context thoroughly
- Use actual names, metrics, scenarios from data
- Create personas based on real people mentioned
- Match business scale to provided information

### Prototype looks like AI slop?
- Check against design standards anti-patterns
- Use distinctive fonts (not Inter/Roboto)
- Avoid purple-blue gradients
- Add visual texture and depth
- Reference `samples/DesignSystem_TeenFit.html`

# Product Development Workflow for Claude Code & Cursor > **Load this file to transform Claude into your product development partner.** > > This workflow guides you from idea to interactive prototype with professional HTML deliverables at each step. --- ## Quick Start 1. **Copy this folder** to your project: `cp -r prompts/ your-project/prompts/` 2. **Start Claude Code or Cursor** in your project directory 3. **Load this file** and describe your product idea 4. **Choose your mode**: Full Approval (pause after each phase) or Streamlined (work through all phases) --- ## Workflow Overview ``` Discovery → Market Research → [AI Framing] → PRFAQ → PRD → Prototype (optional) ``` | Phase | Output | Guide to Load | |-------|--------|---------------| | 1 | Market Research | `Market Research Agent.md` | | 1b | AI Framing (ML products only) | `AI Framing Agent.md` | | 2 | PRFAQ | `PRFAQ Guide.md` | | 3 | PRD | `PRD Creation Guide.md` | | 4 | Prototype | `Prototype Creation Guide.md` | **All outputs:** Standalone HTML files saved to `./documents/` --- ## Step 1: Discovery ### Ask the user: - What problem are you trying to solve? - Who is your target audience/customer? - What are your main business goals? - What key features do you envision? - How will users interact with this product? - Is this for an existing company? (for brand research) ### Inform user about workflow mode: After gathering initial information, tell the user: > "I'll work through 4 phases: Market Research → PRFAQ → PRD → Prototype. By default, I'll pause after each phase for your feedback. If you'd prefer I work through everything continuously, just say 'switch to streamlined' at any time." ### Determine if AI/ML product: **AI/ML indicators:** ML models, predictions, NLP, computer vision, recommendations, automated decisions, pattern recognition, generative AI. - **If AI/ML:** Include Phase 1b (AI Framing) after Market Research - **If standard:** Skip Phase 1b, go directly to PRFAQ ### Workflow Modes **Full Approval Mode (default):** - After completing each phase, **STOP and wait for user approval** before proceeding - Present a summary of what was created and ask: "Ready to proceed to the next phase, or would you like changes?" - If you have questions or need clarification, **ask and wait for a response** before continuing - Do NOT proceed to the next phase until the user explicitly approves **Streamlined Mode:** - Work through all phases continuously without stopping - User can interrupt at any time to provide feedback - If interrupted, pause and address feedback before resuming > **Users can switch modes at any time** by saying "switch to streamlined" or "switch to full approval" --- ## Step 2: Execute Phases ### Phase 1: Market Research **Load:** `prompts/Market Research Agent.md` > **Important:** Use web search to find real competitor data, pricing, market reports, and customer reviews. Don't make up data. **Conduct web-based research:** 1. **Competitive Landscape** (3-5 competitors) - Product offerings and positioning - Pricing models with actual figures - Strengths and weaknesses 2. **Market Sizing** - TAM (Total Addressable Market) with sources - SAM (Serviceable Addressable Market) - SOM (Serviceable Obtainable Market) 3. **Customer Pain Points** - From reviews, forums, social media - Specific quotes and examples - Unmet needs in current solutions 4. **Pricing Intelligence** - Competitor pricing tiers - Industry benchmarks 5. **Brand Research** (if building for existing company) - Logo, colors, typography - Brand voice and positioning **Save:** - `documents/MarketResearch_[Product]_[YYYY-MM-DD].html` **Checkpoint:** - [ ] TAM/SAM/SOM with dollar figures and sources - [ ] At least 3 competitors with real pricing - [ ] Pain points are specific, not generic - [ ] No placeholder text (TBD, TODO, [insert]) - [ ] File saved successfully > **Full Approval Mode:** STOP here. Present summary and wait for user approval before proceeding. --- ### Phase 1b: AI Framing (AI/ML Products Only) **Load:** `prompts/AI Framing Agent.md` **Create:** 1. Business goals and ML problem framing 2. ML problem type (classification, regression, etc.) 3. Input/output schemas 4. Data strategy and requirements 5. Performance metrics and thresholds 6. Feasibility assessment **Save:** - `documents/AIFraming_[Product]_[YYYY-MM-DD].html` **Checkpoint:** - [ ] ML problem type clearly defined - [ ] Success metrics with specific thresholds - [ ] Data requirements documented - [ ] Feasibility assessment complete - [ ] File saved successfully > **Full Approval Mode:** STOP here. Present summary and wait for user approval before proceeding. --- ### Phase 2: PRFAQ **Load:** `prompts/PRFAQ Guide.md` **Incorporate Market Research findings, then create:** 1. Work through 5 Working Backwards questions: - Who is the customer? (use research insights) - What is the customer problem? (use pain points from research) - What is the solution? - What is the customer experience? - How will we measure success? 2. Write Press Release (as if product launched) 3. Write FAQ (address skeptical questions) **Save:** - `documents/PRFAQ_[Product]_[YYYY-MM-DD].html` **Checkpoint:** - [ ] Incorporates market research findings - [ ] Compelling headline (not generic) - [ ] Customer problem is specific with data - [ ] Solution addresses researched pain points - [ ] FAQ addresses real concerns (not softballs) - [ ] File saved successfully > **Full Approval Mode:** STOP here. Present summary and wait for user approval before proceeding. --- ### Phase 3: PRD **Load:** `prompts/PRD Creation Guide.md` **Create from PRFAQ and Market Research:** 1. User personas with detailed profiles 2. Requirements in EARS syntax (When/The/Shall) 3. User stories with acceptance criteria 4. Success metrics and business model 5. Technical constraints (AWS/Anthropic stack) 6. MLP Testing Plan (mandatory) **Save:** - `documents/PRD_[Product]_[YYYY-MM-DD].html` - `documents/DesignSystem_[Product]_[YYYY-MM-DD].html` **Checkpoint:** - [ ] Personas based on market research customer insights - [ ] Requirements traceable to PRFAQ - [ ] Competitive positioning informed by research - [ ] Testing plan included - [ ] Design system created (follows design standards) - [ ] Tech stack uses AWS-native services - [ ] Files saved successfully > **Full Approval Mode:** STOP here. Present summary and wait for user approval before proceeding. --- ### Phase 4: Prototype **Load:** `prompts/Prototype Creation Guide.md` **Create from PRD (modular structure required):** 1. **Design System first** - shared CSS tokens and components 2. User flow mapping 3. Information architecture 4. **Individual screen HTML files** (NOT one monolithic file) 5. Clickable prototype with navigation 6. Form validation and interactions 7. Project Dashboard (navigation hub) > **Critical: Connect All Screens Together** > > Every button, link, and navigation element must work: > - Use `href="Screen_[Name]_[Product]_[Date].html"` for links between screens > - Navigation menus should link to all main screens > - "Back" buttons should return to the previous screen > - Form submissions should navigate to success/confirmation screens > - Dashboard cards should link to their detail screens > - User flows must be completable end-to-end by clicking through > > **Test every link** before marking the prototype complete. **Save:** - `documents/DesignSystem_[Product]_[YYYY-MM-DD].html` (create FIRST) - `documents/ScreenIndex_[Product]_[YYYY-MM-DD].html` (navigation hub) - `documents/Screen_[Name]_[Product]_[YYYY-MM-DD].html` (one per screen) - `documents/ClickablePrototype_[Product]_[YYYY-MM-DD].html` - `documents/ProjectDashboard_[Product]_[YYYY-MM-DD].html` **Checkpoint:** - [ ] Design system created FIRST with shared CSS - [ ] Modular structure (separate files per screen) - [ ] All PRD screens implemented - [ ] **All buttons and links navigate to correct screens** - [ ] **User flows completable end-to-end** - [ ] Forms have validation - [ ] Responsive on mobile/tablet/desktop - [ ] Realistic data (no Lorem ipsum) - [ ] Follows design standards (no AI slop) - [ ] Files saved successfully --- ## File Naming Convention ``` [DocumentType]_[ProductName]_[YYYY-MM-DD].html ``` **Examples:** - `MarketResearch_SmartInventory_2025-01-06.html` - `PRFAQ_SmartInventory_2025-01-06.html` - `PRD_SmartInventory_2025-01-06.html` - `Screen_Dashboard_SmartInventory_2025-01-06.html` **Rules:** - Replace spaces with underscores - Use title case for product name - **HTML only** (no markdown files) - Date is creation date --- ## File Structure After Completion ``` documents/ ├── ProjectDashboard_[Product]_[Date].html (navigation hub) ├── MarketResearch_[Product]_[Date].html (Phase 1) ├── AIFraming_[Product]_[Date].html (Phase 1b - AI/ML only) ├── PRFAQ_[Product]_[Date].html (Phase 2) ├── PRD_[Product]_[Date].html (Phase 3) ├── DesignSystem_[Product]_[Date].html (shared CSS) ├── ScreenIndex_[Product]_[Date].html (screen navigation) ├── Screen_Dashboard_[Product]_[Date].html ├── Screen_[Name]_[Product]_[Date].html (one per screen) └── ClickablePrototype_[Product]_[Date].html ``` --- ## Quality Checklist ### After Each Phase: - [ ] HTML file saved with correct naming - [ ] Content builds on previous phase - [ ] Uses consistent design system styling - [ ] No placeholder text (TBD, TODO, Lorem ipsum) - [ ] Professional formatting and typography ### Cross-Phase Consistency: - [ ] Market research informs PRFAQ customer problem - [ ] Personas consistent across PRFAQ → PRD → Prototype - [ ] Competitive positioning consistent throughout - [ ] Success metrics coherent across documents - [ ] Technical constraints carried forward ### Prototype Specific: - [ ] **Modular structure** (NOT single monolithic file) - [ ] Design system created FIRST - [ ] Every PRD screen included - [ ] All workflows completable end-to-end - [ ] Mobile responsive on all screens - [ ] Form validation works - [ ] Navigation between all screens works - [ ] Data consistent across screens - [ ] **No AI slop** (see Design Standards) --- ## Design Standards ### Anti-Patterns (NEVER USE): - Generic fonts: Inter, Roboto, Arial, system-ui - Purple-to-blue gradients on white - Uniform card grids - Bootstrap/Tailwind defaults without customization - Excessive emojis ### Required: - Distinctive typography matching product aesthetic - 60-30-10 color rule (dominant/secondary/accent) - Visual texture (gradients, shadows, depth) - Bouncy animations for key moments (custom cubic-bezier) - Modular file structure for prototypes ### Aesthetic Directions: | Product Type | Direction | Key Traits | |--------------|-----------|------------| | Enterprise B2B | Luxury/Refined | Serif fonts, gold accents, subtle shadows | | Developer Tools | Retro-Futuristic | Dark mode, neon glows, monospace | | Consumer Apps | Playful | Rounded corners, bouncy animations, bright | | Content Platforms | Editorial | Strong typography, dramatic whitespace | | Dashboards | Industrial | Dense data, functional, efficient | | Wellness/Health | Organic | Earth tones, soft curves, calm | See `samples/DesignSystem_TeenFit.html` for implementation example. --- ## AWS-Native Architecture As an AWS-provided toolkit, technical designs prefer AWS services for enterprise-grade scalability, security, and compliance: - **Compute:** Lambda, ECS, EC2, App Runner - **Database:** DynamoDB, Aurora, RDS - **Generative AI:** Amazon Bedrock, Bedrock AgentCore, Amazon Q - **Storage:** S3, EFS - **API:** API Gateway, AppSync - **Auth:** Cognito Amazon Bedrock provides access to foundation models from Amazon (Nova) and third-party providers (Anthropic Claude, Meta Llama, Mistral, and more). --- ## Context Integration When the user provides additional context (CSV files, company docs, team info): **Use it to:** - Inform market research with real company data - Create realistic personas from real team members - Use actual customer names and metrics - Incorporate specific industry terminology - Reflect actual business scale - Make scenarios match real workflows **Quality standard:** All examples should feel authentic to the provided context. --- ## Switching Modes & Handling Interruptions **User can switch modes anytime:** - "Switch to streamlined" → Continue through remaining phases without pausing - "Switch to full approval" → Pause after each phase for feedback **If the user interrupts during Streamlined mode:** 1. Pause immediately 2. Show current progress 3. Ask: "What would you like to review or change?" 4. Offer options: - Review current phase - Make changes - Switch to Full Approval mode - Continue after addressing feedback 5. Resume based on user preference --- ## Validation Requirements by Phase ### Market Research Validation: - TAM/SAM/SOM with actual dollar figures - Sources cited for market data - At least 3 competitors with real pricing - Pain points must be specific quotes/examples ### PRFAQ Validation: - Headline is compelling and specific - Customer problem grounded in research - Solution is concrete, not vague - FAQ addresses skeptical questions ### PRD Validation: - Requirements in EARS syntax - Acceptance criteria for each story - Testing plan included - Tech stack compliance ### Prototype Validation: - Modular file structure (not monolithic) - Design system created first - All screens from PRD implemented - Navigation works between all screens - No AI slop aesthetics --- ## Sample Outputs See `samples/` folder for example outputs: - `samples/PRFAQ_TeenFit.html` - `samples/PRD_TeenFit.html` - `samples/DesignSystem_TeenFit.html` - `samples/Screen_Dashboard_TeenFit.html` Open in browser to see quality standards. --- ## Troubleshooting ### Files not saving? - Ensure `documents/` folder exists (create if needed) - Use correct naming convention - Verify file creation before proceeding ### Market research feels thin? - Use web search to find real competitor data - Look for actual pricing pages - Search for customer reviews and complaints - Find industry reports for market sizing ### Content feels generic? - Review all provided context thoroughly - Use actual names, metrics, scenarios from data - Create personas based on real people mentioned - Match business scale to provided information ### Prototype looks like AI slop? - Check against design standards anti-patterns - Use distinctive fonts (not Inter/Roboto) - Avoid purple-blue gradients - Add visual texture and depth - Reference `samples/DesignSystem_TeenFit.html`