Mini-CoderBrain - Quick Reference

One-Page Cheat Sheet Print this for quick reference!

🎭 Behavior Profiles

Profile Selection

# CLAUDE.md (line ~41):
behavior_profile: "default"  # default | focus | research | implementation

Profile Comparison

Profile Tokens Best For Output Suggestions
default 200 General dev ~150 words 2-5/session
focus 150 Bug fixes ~50 words 0-1/session
research 300 Learning ~300 words 5-10/session
implementation 200 Features ~150 words 3-6/session

Apply changes: Save CLAUDE.md + Restart Claude Code


πŸ“š Patterns Library

5 Core Patterns (4,700 lines, zero token cost)

Pattern Purpose Lines
pre-response-protocol Mandatory checklist ~850
context-utilization Memory bank usage ~750
proactive-behavior When to suggest ~800
anti-patterns What NOT to do ~1200
tool-selection-rules Which tool when ~1100

Location: .claude/patterns/*.md Read: On-demand (automatic) Cost: 0 tokens (never injected)


πŸ“Š Metrics Commands

/metrics              # Current session
/metrics --weekly     # Last 7 days
/metrics --profile    # Profile comparison
/metrics --all        # Full report

What’s tracked: Tool usage, duration, profile, operations What’s NOT: Code content, file names, personal data Storage: .claude/metrics/sessions/*.json Retention: 30 days (auto-archived)


πŸš€ Quick Start (v2.1)

1. Verify v2.1

🧠 [MINI-CODERBRAIN: ACTIVE] - ProjectName
🎯 Focus: General Development
πŸ“‚ Context: .claude/memory/ (loaded)
🎭 Profile: default                    ← v2.1 feature
⚑ Ready for development

2. Try Different Profiles

behavior_profile: "focus"     # Deep concentration
behavior_profile: "research"  # Detailed learning
behavior_profile: "implementation" # Rapid building

3. Check Metrics

/metrics --weekly    # Review your patterns

πŸ”§ Custom Profile Template

# .claude/profiles/my-profile.md

## Core Settings
output_style: terse              # terse | balanced | verbose
proactivity: low                 # low | medium | high
explanation_depth: minimal       # minimal | standard | detailed
communication: efficient         # efficient | friendly | educational

## Focus Areas
### Primary
- [Your main focus area]

## Tool Preferences
### Prefer
- Read, Edit, Grep

### Avoid
- Task (keep focused)

# Activate in CLAUDE.md:
behavior_profile: "my-profile"

πŸ“ Key Files & Locations

Memory Bank (Context)

.claude/memory/
β”œβ”€β”€ productContext.md        # Project overview
β”œβ”€β”€ systemPatterns.md        # Coding standards
β”œβ”€β”€ activeContext.md         # Current focus
β”œβ”€β”€ progress.md              # Task tracking
└── decisionLog.md           # Technical decisions

Behavioral System (v2.1)

.claude/patterns/            # Behavioral rules (4,700 lines)
.claude/profiles/            # AI modes (4 core + custom)
.claude/metrics/             # Effectiveness tracking

Configuration

CLAUDE.md                    # Main controller
  └── Line ~41: behavior_profile

⚑ Common Tasks

Switch Profile

# 1. Edit CLAUDE.md
behavior_profile: "focus"

# 2. Restart Claude Code

# 3. Verify
# Should show: 🎭 Profile: focus

View Metrics

/metrics                # Quick session view
/metrics --weekly       # Weekly patterns

Create Custom Profile

# 1. Copy template
cp .claude/profiles/custom-template.md .claude/profiles/mine.md

# 2. Edit settings
nano .claude/profiles/mine.md

# 3. Activate
# In CLAUDE.md: behavior_profile: "mine"

Check Context Quality

/validate-context

🎯 Decision Matrix

When to Use Each Profile

Need deep focus?              β†’ focus
Learning new codebase?        β†’ research
Building features fast?       β†’ implementation
General development?          β†’ default
Production hotfix?            β†’ focus
Architecture review?          β†’ research
Sprint work?                  β†’ implementation
Not sure?                     β†’ default

πŸ“Š Token Impact

Component Token Cost Impact
Patterns Library 0 Read on-demand
Profile (focus) ~150 25% lighter
Profile (default) ~200 Baseline
Profile (research) ~300 50% heavier
Metrics System 0 Background
Total v2.1 +200 avg ~0.1% increase

v2.0: 79.9% reduction v2.1: 79.8% reduction Result: Virtually unchanged efficiency!


πŸ› Quick Troubleshooting

Profile Not Showing?

# Check setting
grep "behavior_profile" CLAUDE.md

# Verify file exists
ls .claude/profiles/default.md

# Restart Claude Code

Metrics No Data?

# Install jq
sudo apt-get install jq  # Ubuntu
brew install jq          # macOS

# Wait for session end
# Then try: /metrics

Different Behavior?

# Reset to v2.0 behavior
behavior_profile: "default"

# Restart Claude Code

Getting Started:

User Guides:

Reference:


πŸ’‘ Pro Tips

1. Profile Rotation

Morning:   research   (learn)
Afternoon: implementation (build)
Evening:   focus   (fix bugs)

2. Metrics-Driven Selection

# After 2 weeks:
/metrics --profile

# See which profile works best for you
# Adjust accordingly

3. Custom Task Profiles

Create:
- backend-focused.md
- frontend-focused.md
- devops-focused.md

Match profile to recurring workflows

4. Weekly Review

# Every Monday:
/metrics --weekly

# Identify patterns
# Optimize workflow

⚑ Quick Command Reference

# Core Commands
/init-memory-bank        # Initialize (mandatory first run)
/update-memory-bank      # Update context
/validate-context        # Check quality
/map-codebase           # Index files

# v2.1 Commands
/metrics                 # Current session
/metrics --weekly        # Last 7 days
/metrics --profile       # Compare profiles
/metrics --all           # Full report

# Memory Management
/memory-sync            # Sync memory bank
/memory-cleanup         # Archive old data
/context-update         # Quick context update

πŸ”„ Backward Compatibility

v2.1 is 100% backward compatible with v2.0

  • Default profile = v2.0 behavior
  • No breaking changes
  • All v2.0 features work
  • Optional feature activation

Upgrade risk: None Migration time: 5 minutes


πŸ“ž Get Help

Issues: GitHub Issues Docs: GitHub Pages Version: 2.1.0 Status: Production-ready


Print this page for quick reference! πŸ“„


Mini-CoderBrain v2.1 - Data-driven behavioral intelligence