Skip to content

MIGRATION-V2.1.md

Latest commit

 

History

History
368 lines (261 loc) · 9.19 KB

MIGRATION-V2.1.md

File metadata and controls

368 lines (261 loc) · 9.19 KB
Raw

Migration Guide: v2.0 → v2.1

Version: 2.0.0 → 2.1.0 Release Date: October 2025 Migration Time: 5 minutes Breaking Changes: None (100% backward compatible)

🎯 TL;DR - Quick Migration

Good News: v2.1 is 100% backward compatible! Your v2.0 project works unchanged.

To upgrade:

  1. Pull latest changes from v2.1-development branch
  2. Restart Claude Code
  3. (Optional) Try new profiles: edit CLAUDE.md line ~41

That's it! v2.1 features are automatically available.

✨ What's New in v2.1

1. Behavior Profiles (NEW)

Customizable AI modes for different development contexts.

4 Core Profiles:

  • default (200 tokens) - Balanced mode (same as v2.0 behavior)
  • focus (150 tokens) - Deep concentration, minimal output
  • research (300 tokens) - Detailed exploration and learning
  • implementation (200 tokens) - Rapid feature building

Plus: Custom profile template for creating your own

2. Behavioral Patterns Library (NEW)

4,700 lines of behavioral training, accessible on-demand.

5 Core Patterns:

  • pre-response-protocol.md - MANDATORY 5-step checklist
  • context-utilization.md - Memory bank usage without duplication
  • proactive-behavior.md - When/how to suggest things
  • anti-patterns.md - What NOT to do (1200 lines of examples)
  • tool-selection-rules.md - Which tool for each task

Token Cost: +0 (patterns read on-demand, not injected)

3. Smart Metrics System (NEW)

Automatic tracking of behavioral effectiveness.

Features:

  • Background metrics collection (zero user action)
  • /metrics command for insights
  • Tool usage tracking
  • Profile performance comparison
  • Privacy-first (no sensitive data stored)

Token Cost: +0 (background collection only)

📊 Impact Comparison

Feature v2.0 v2.1 Change
Token Efficiency 79.9% reduction 79.8% reduction -0.1% (minimal)
Conversation Length 100+ turns 100+ turns Same
Behavioral Modes 1 (fixed) 4+ (customizable) NEW
Patterns Library Inline Modular (4,700 lines) NEW
Metrics Tracking None Automatic NEW
Profiles None 4 + custom NEW

Net Token Impact: +200 tokens per session (~0.1% of context window)

🔄 Migration Steps

Step 1: Update Your Code

Option A: Git Pull (Recommended)

cd /path/to/your/project
git fetch origin
git checkout v2.1-development
git pull origin v2.1-development

Option B: Manual Update

# Backup your current .claude/ folder
cp -r .claude .claude-backup

# Download v2.1
git clone -b v2.1-development https://github.com/kunwar-shah/mini-coder-brain.git mini-coder-brain-v2.1
cd mini-coder-brain-v2.1

# Copy new files to your project
cp -r .claude/patterns /path/to/your/project/.claude/
cp -r .claude/profiles /path/to/your/project/.claude/
cp -r .claude/metrics /path/to/your/project/.claude/
cp .claude/hooks/metrics-collector.sh /path/to/your/project/.claude/hooks/
cp .claude/hooks/metrics-report.sh /path/to/your/project/.claude/hooks/
cp .claude/commands/metrics.md /path/to/your/project/.claude/commands/

# Update modified files
cp .claude/hooks/session-start.sh /path/to/your/project/.claude/hooks/
cp .claude/hooks/session-end.sh /path/to/your/project/.claude/hooks/
cp CLAUDE.md /path/to/your/project/

# Make scripts executable
chmod +x /path/to/your/project/.claude/hooks/*.sh

Step 2: Restart Claude Code

Close and reopen Claude Code in your project.

Verify v2.1 is active:

🧠 [MINI-CODERBRAIN: ACTIVE] - YourProject
🎯 Focus: General Development
📂 Context: .claude/memory/ (loaded)
🎭 Profile: default                    ← NEW LINE
⚡ Ready for development

If you see 🎭 Profile: default, v2.1 is working!

Step 3: Optional - Try Behavior Profiles

Edit CLAUDE.md (around line 41):

# Behavior Profile (v2.1+)
behavior_profile: "focus"  # Try: default, focus, research, implementation

Save and restart Claude Code to see the new profile in action.

🎭 Understanding Profiles

default (200 tokens)

Same as v2.0 behavior - Balanced, context-aware, proactive.

Use when:

  • General full-stack development
  • Learning new project
  • Unsure which profile to use

Output: Standard explanations, 2-5 suggestions per session

focus (150 tokens)

Deep concentration mode - Terse, minimal output, low proactivity.

Use when:

  • Complex bug fixes
  • Time-sensitive delivery
  • Need maximum concentration

Output: ~50 word responses, 0-1 suggestions per session Token Savings: 25% lighter than default

research (300 tokens)

Exploration mode - Verbose, detailed explanations, high proactivity.

Use when:

  • Understanding new codebase
  • Learning patterns
  • Creating documentation

Output: ~300 word responses, 5-10 suggestions per session Token Cost: 50% heavier than default (worth it for learning)

implementation (200 tokens)

Rapid build mode - Action-oriented, step-by-step execution.

Use when:

  • Building new features
  • Prototyping MVP
  • Sprint work

Output: Step-by-step progress, 3-6 suggestions per session

📊 Using Metrics

View Current Session

/metrics

Output:

📊 METRICS - Current Session (45 mins, default profile)

Tool Usage:
Read: 12 | Edit: 8 | Bash: 4 | Grep: 5

Session Outcome:
✅ Tasks Completed: 3
⚠️  Errors: 1

View Weekly Summary

/metrics --weekly

Compare Profiles

/metrics --profile

Insights:

  • Which tools you use most
  • Which profiles are most effective
  • Session duration patterns
  • Task completion rates

🔧 Advanced: Custom Profiles

Create Your Own Profile

  1. Copy template:

    cp .claude/profiles/custom-template.md .claude/profiles/my-workflow.md

  2. Edit settings:

    ## Core Settings
output_style: terse           # terse, balanced, verbose
proactivity: low              # low, medium, high
explanation_depth: minimal    # minimal, standard, detailed

  3. Activate:

    # In CLAUDE.md:
behavior_profile: "my-workflow"

See .claude/profiles/custom-template.md for full template.

🐛 Troubleshooting

Profile Not Showing

Problem: Session status doesn't show 🎭 Profile: ...

Solution:

  1. Check CLAUDE.md line ~41 has behavior_profile: setting
  2. Restart Claude Code
  3. If still missing, profile detection may have failed - defaults to "default" profile

Metrics Not Working

Problem: /metrics command shows "No session data"

Solution:

  1. Metrics collected at session end - work for a bit first
  2. Check .claude/metrics/sessions/ folder exists
  3. Ensure jq is installed: apt-get install jq (or brew install jq on macOS)

Different Behavior Than v2.0

Problem: Claude seems to behave differently

Reason: You may have accidentally changed profile

Solution:

  1. Check CLAUDE.md line ~41
  2. Set to behavior_profile: "default" for v2.0 behavior
  3. Restart Claude Code

↩️ Rolling Back to v2.0

If you encounter issues, easily rollback:

Option 1: Git Tag

git checkout v2.0-stable

Option 2: Restore Backup

cp -r .claude-backup .claude

Option 3: Reinstall v2.0

git clone -b main https://github.com/kunwar-shah/mini-coder-brain.git
# Follow v2.0 installation steps

📚 Additional Resources

Documentation

Customization

Support

✅ Migration Checklist

Before considering migration complete:

  • Pulled v2.1 code or manually updated files
  • Restarted Claude Code
  • Verified 🎭 Profile: default appears in session status
  • (Optional) Tried different profiles
  • (Optional) Ran /metrics command
  • Confirmed project works as expected
  • (Optional) Created custom profile for your workflow

🎉 Welcome to v2.1!

You now have:

  • ✅ Customizable AI behavior via profiles
  • ✅ 4,700 lines of training accessible on-demand
  • ✅ Automatic metrics tracking for data-driven insights
  • ✅ 100% backward compatibility with v2.0
  • ✅ Same token efficiency (79.8% reduction maintained)

Enjoy your enhanced Mini-CoderBrain experience! 🚀

Questions? Open an issue on GitHub