Mini-CoderBrain - Quick Reference
One-Page Cheat Sheet | Print this for quick reference! |
π 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
π 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)
π Metrics Commands
/metrics # Current session
/metrics --weekly # Last 7 days
/metrics --profile # Profile comparison
/metrics --all # Full report
Whatβs tracked: Tool usage, duration, profile, operations
Whatβs NOT: Code content, file names, personal data
Storage: .claude/metrics/sessions/*.json
Retention: 30 days (auto-archived)
π Quick Start (v2.1)
1. Verify v2.1
π§ [MINI-CODERBRAIN: ACTIVE] - ProjectName
π― Focus: General Development
π Context: .claude/memory/ (loaded)
π Profile: default β v2.1 feature
β‘ 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
π§ 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"
π Key Files & Locations
Memory Bank (Context)
.claude/memory/
βββ productContext.md # Project overview
βββ systemPatterns.md # Coding standards
βββ activeContext.md # Current focus
βββ progress.md # Task tracking
βββ 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
βββ Line ~41: behavior_profile
β‘ Common Tasks
Switch Profile
# 1. Edit CLAUDE.md
behavior_profile: "focus"
# 2. Restart Claude Code
# 3. Verify
# Should show: π 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
π― 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
π 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!
π 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
π Documentation Links
Getting Started:
- README - Project overview
- Migration v2.0βv2.1 - Upgrade guide
User Guides:
- Profiles Guide - Complete profiles documentation
- Patterns Guide - Behavioral patterns reference
- Metrics Guide - Metrics system guide
Reference:
- FAQ - Common questions
- Release Notes - Whatβs new
π‘ 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
β‘ 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
π 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
π Get Help
Issues: GitHub Issues Docs: GitHub Pages Version: 2.1.0 Status: Production-ready
Print this page for quick reference! π
Mini-CoderBrain v2.1 - Data-driven behavioral intelligence