Skip to content

setup-guide.md

Latest commit

 

History

History
396 lines (291 loc) · 12.6 KB

setup-guide.md

File metadata and controls

396 lines (291 loc) · 12.6 KB
Raw

ADD Framework Integration - Setup Guide

Complete step-by-step instructions for integrating the Assess-Decide-Do framework into your Claude workflow.

Prerequisites

  • Access to Claude (via claude.ai, Claude Desktop, or Claude Code)
  • Basic familiarity with markdown files
  • (Optional) Git for cloning the repository

Installation Options

Choose the integration level that matches your needs:

Option 0: Skills-First for Claude Code (10 minutes) (Recommended)

Best for: Claude Code users wanting efficient, modern integration

What you get:

  • ~1.5KB base context (vs 26KB with megaprompt)
  • Skills load on demand when needed
  • /add-status and /add-reflect commands
  • Automatic flow tracking

Steps:

  1. Clone the Repository

    git clone https://github.com/dragosroua/claude-assess-decide-do-mega-prompt.git
cd claude-assess-decide-do-mega-prompt

  2. Open in Claude Code The integration activates automatically:

    • CLAUDE.md provides core ADD awareness
    • .claude/skills/ contains all ADD skills
    • .claude/settings.json configures the statusline

  3. Test the Skills

    /add-status    # Shows current realm and metrics
/add-reflect   # Captures session patterns to history

  4. Watch the Statusline The bottom of Claude Code shows real-time realm:

    [ADD Flow: 🔴+ Assess | Exploring options]

  5. Verify Success

    • Responses align with your realm
    • Status updates at conversation boundaries
    • /add-status shows meaningful metrics

For your own projects: Copy .claude/ directory and CLAUDE.md to your project root.

Option 1: Quick Start - Custom Instructions (5 minutes)

Best for: claude.ai users who want ADD awareness in all conversations (no skills support)

Steps:

  1. Access Claude Settings

    • Go to claude.ai and log in
    • Click your profile icon (bottom left)
    • Select "Settings"
    • Navigate to "Custom Instructions"

  2. Add ADD Framework Instructions Copy and paste this into your custom instructions:

    Framework: Operate with Assess-Decide-Do (ADD) life management framework.

Core principles:
- Detect user's current realm: Assess (explore), Decide (commit), Do (execute)
- Identify imbalances: analysis paralysis, decision avoidance, execution shortcuts
- Guide balanced flow between realms
- Structure responses appropriately for each realm

When relevant, reference ADD_FRAMEWORK_MEGAPROMPT.md for detailed guidance.

  3. Test the Integration Start a new conversation and try:

    • "I'm exploring options for X..." (Assess realm - Claude should support exploration)
    • "I need to choose between A and B..." (Decide realm - Claude should support decision-making)
    • "How do I implement X?" (Do realm - Claude should provide execution guidance)

  4. Verify Success You should notice:

    • Responses aligned with your realm
    • Gentle guidance when stuck in one realm
    • No premature pushing (e.g., to decide when you're assessing)

Option 2: Per-Conversation Activation (2 minutes per conversation)

Best for: Users who want ADD awareness for specific conversations only

Steps:

  1. Clone or Download This Repository

    git clone https://github.com/dragosroua/claude-assess-decide-do-mega-prompt.git

    Or download as ZIP from GitHub

  2. Choose and Locate Your Mega Prompt

    Option A: Generic Version (recommended for most users)

    • File: docs/ADD_FRAMEWORK_MEGAPROMPT.md
    • Clean ADD framework without user-specific context

    Option B: Personalized Version (advanced)

    • File: docs/ADD_FRAMEWORK_MEGAPROMPT_USER_CONTEXT.md
    • Contains user context section
    • Edit the "INTEGRATION WITH USER CONTEXT" section with your own background
    • Useful if you want Claude to understand your specific relationship with ADD

  3. Start Your Conversation Begin any Claude conversation with:

    Load and operate with ADD framework awareness from ADD_FRAMEWORK_MEGAPROMPT.md
throughout this conversation. Detect realms, identify imbalances, guide
balanced flow.

  4. Upload the Mega Prompt (if using web interface)

    • Attach docs/ADD_FRAMEWORK_MEGAPROMPT.md to the conversation
    • Claude will read and integrate it automatically

  5. For Claude Desktop/Code

    • Place the repository in an accessible location
    • Reference the file path in your initial message

Option 3: Project-Level Integration - Megaprompt (10 minutes)

Best for: Claude Desktop users or projects where skills aren't supported

Note: For Claude Code, use Option 0 (Skills-First) instead—it's more efficient.

Steps:

  1. Clone the Repository

    git clone https://github.com/dragosroua/claude-assess-decide-do-mega-prompt.git
cd claude-assess-decide-do-mega-prompt

  2. Copy Files to Your Project

    # Create ADD directory in your project
mkdir -p your-project/.claude-add

# Copy essential files
# Choose generic version OR personalized version (not both)
cp docs/ADD_FRAMEWORK_MEGAPROMPT.md your-project/.claude-add/
# OR for personalized version (edit USER CONTEXT section first):
# cp docs/ADD_FRAMEWORK_MEGAPROMPT_USER_CONTEXT.md your-project/.claude-add/ADD_FRAMEWORK_MEGAPROMPT.md

cp docs/ADD_QUICK_REFERENCE.md your-project/.claude-add/

  3. Create .claude File In your project root, create .claude:

    instructions: |
  Operate with Assess-Decide-Do (ADD) life management framework.
  
  Core integration:
  - Detect user's realm (Assess/Decide/Do) from language patterns
  - Identify workflow imbalances
  - Structure responses appropriate to current realm
  - Guide balanced flow without being heavy-handed
  
  Load ADD_FRAMEWORK_MEGAPROMPT.md for complete framework guidance.

context_files:
  - .claude-add/ADD_FRAMEWORK_MEGAPROMPT.md
  - .claude-add/ADD_QUICK_REFERENCE.md

  4. Test in Claude Code

    • Open your project in Claude Code
    • Start a coding session
    • Claude should operate with ADD awareness automatically

  5. Verify Integration Try these tests:

    • Explore technical approaches (Assess) - no premature implementation
    • Discuss architecture decisions (Decide) - supportive of commitment
    • Request implementation (Do) - focused execution guidance

Option 4: Deep Integration with Memory (Advanced)

Best for: Users wanting persistent ADD awareness across all sessions

Steps:

  1. Complete Option 1 (Custom Instructions) Follow steps above to set base integration

  2. Add Memory Edits In any Claude conversation, say:

    Please add these memory edits:
1. "User operates with Assess-Decide-Do (ADD) framework"
2. "Detect user's realm and guide balanced flow"
3. "Reference ADD_FRAMEWORK_MEGAPROMPT.md for ADD integration"

  3. Verify Memory Integration Start a new conversation and check if Claude:

    • References ADD framework naturally
    • Detects realms without being told
    • Guides flow appropriately

  4. Calibrate Over Time As you use ADD-enhanced Claude:

    • Provide feedback on interventions (too pushy vs. too subtle)
    • Refine memory edits based on what works
    • Adjust custom instructions as needed

Configuration Examples

Example 1: Content Creator Setup

Custom Instructions:

ADD Framework for Content Creation:
- Assess: Brainstorm topics, research, explore angles
- Decide: Choose structure, commit to key messages
- Do: Write, edit, publish

Detect realm from my language. Guide balanced flow.
Load ADD_FRAMEWORK_MEGAPROMPT.md when needed.

Example 2: Developer Setup

.claude file:

project: "My Development Project"

instructions: |
  ADD Framework for Development:
  - Assess: Explore technical approaches, evaluate trade-offs
  - Decide: Commit to architecture, choose implementation
  - Do: Code, test, deploy
  
  Enforce realm boundaries:
  - No premature implementation during assessment
  - No re-architecture during execution
  
  Load ADD_FRAMEWORK_MEGAPROMPT.md for complete guidance.

context_files:
  - .claude-add/ADD_FRAMEWORK_MEGAPROMPT.md

Example 3: Business Strategist Setup

Custom Instructions + Memory:

Custom Instructions:
ADD Framework for Business Strategy
- Assess realm: Market research, competitive analysis
- Decide realm: Focus areas, resource allocation
- Do realm: Campaign execution, measurement

Memory Edits:
1. "User applies ADD framework to business strategy"
2. "Detect realm from business language patterns"
3. "Guide flow between research, decision, and execution phases"

Verification & Testing

Quick Verification Test

Run this test sequence in a new conversation:

Test 1: Assess Detection

You: "I'm exploring options for a new blog. What topics could I write about?"

Expected: Claude provides expansive exploration without pushing to decide.
Should NOT say: "You should choose X" or "Start with Y"

Test 2: Decide Detection

You: "I've explored 5 blog topics thoroughly. I need to choose one. Help me decide."

Expected: Claude supports decision-making, asks values-based questions.
Should NOT say: "Let me give you more options to explore"

Test 3: Do Detection

You: "I've decided on the blog topic. How do I actually start writing?"

Expected: Claude provides clear execution steps.
Should NOT say: "Are you sure about that topic? Maybe reconsider..."

Test 4: Imbalance Detection

You: [After 5 messages exploring] "Maybe I should research 10 more topics first?"

Expected: Claude gently identifies analysis paralysis and guides toward Decide.
Should say something like: "You've done thorough assessment. Ready to explore what choosing might look like?"

Full Test Scenarios

See examples/test-scenarios.md for comprehensive testing instructions.

Troubleshooting Common Issues

Issue: "Claude doesn't seem ADD-aware"

Solutions:

  1. Verify mega prompt is loaded (check file path)
  2. Ensure custom instructions are saved
  3. Try explicit activation: "Operate with ADD framework from now on"
  4. Check if you're using clear realm language in your requests

Issue: "Interventions feel too heavy-handed"

Solutions:

  1. Provide calibration feedback: "That felt pushy"
  2. Add to instructions: "Guide subtly, don't lecture about ADD"
  3. Request: "Operate with ADD but keep framework mostly invisible"

Issue: "Not noticing any difference"

Solutions:

  1. Test with deliberate imbalances (stay in Assess for extended period)
  2. Use explicit realm language ("I'm exploring..." vs "I need to decide...")
  3. Try comparison: same task with and without ADD integration
  4. Focus on feeling/flow rather than explicit framework mentions

Issue: "Claude mentions ADD too much"

Solutions:

  1. Say: "Operate with ADD awareness but don't name it unless relevant"
  2. Update instructions: "Framework operates implicitly, not explicitly"
  3. Provide feedback: "Less ADD terminology, more natural guidance"

Success Indicators

You'll know integration is working when you experience:

✅ Reduced friction - Fewer misaligned responses ✅ Clearer thinking - Better workflow structure ✅ Natural flow - Smooth realm transitions ✅ Balanced work - Not stuck in one realm ✅ Better completions - Higher finish rate ✅ Relational quality - Claude feels more collaborative

Next Steps

After successful integration:

  1. Use Daily - Let ADD awareness become natural
  2. Provide Feedback - Help calibrate interventions
  3. Explore Domains - Try ADD in different work contexts
  4. Share Insights - Contribute to GitHub discussions
  5. Evolve Setup - Refine based on what works for you

Support & Resources

  • GitHub Issues: Report problems or ask questions
  • Documentation: Read complete guides in docs/
  • Examples: Check examples/ for configuration files
  • Blog Post: Read the full story at dragosroua.com
  • addTaskManager: Try the iOS/macOS app for deeper ADD experience

Advanced Configuration

For advanced users wanting deeper integration:

  • Skills Documentation: See docs/ADD_SKILLS.md for full skills reference
  • Custom Skills: Build ADD-aware skills for specific domains (copy existing skills as templates)
  • MCP Server Integration: See docs/ADD_TECHNICAL_INTEGRATION.md
  • Session History: /add-reflect saves to .add-session-history.md for long-term pattern tracking
  • Statusline Customization: Edit .claude/statusline.sh to change display format

Questions? Open an issue on GitHub or visit dragosroua.com for more ADD resources.

Ready to begin? Pick your integration option above and start experiencing ADD-enhanced Claude collaboration.