How to Use Mini-CoderBrain v2.0

Comprehensive guide for setting up and using mini-coder-brain effectively in any project type.

📋 Table of Contents

1. Quick Start
2. Project Scenarios
3. Essential Commands
4. Best Practices
5. Context Quality
6. Troubleshooting

🚀 Quick Start

Installation + Setup (2 Minutes)

Step 1: Install

git clone https://github.com/kunwar-shah/mini-coder-brain.git
cd mini-coder-brain
./install.sh /path/to/your/project

Step 2: Initialize (MANDATORY)

# In Claude Code, run ONE of these:
/init-memory-bank --docs-path ./docs  # With documentation
/init-memory-bank                     # Auto-detect

Step 3: Verify

/validate-context

Expected: 📊 Context Quality: 85% (Recommended) ✅

🎯 Project Scenarios

Scenario 1: New Empty Project

Context: Starting from scratch, no code yet

Steps:

1. Install mini-coder-brain
2. Run /init-memory-bank
3. Answer interactive prompts:
   • Project name?
   • What type of app?
   • Tech stack preference?
   • Core features?
4. Start building with Claude’s help

Result: 60-70% quality (minimum but functional)

Example:

You: /init-memory-bank

Claude: What is your project name?
You: my-saas-app

Claude: What type of application?
You: A SaaS platform for team collaboration

Claude: What technology stack?
You: Next.js 14, TypeScript, Prisma, PostgreSQL

✅ Context initialized! Quality: 65%

Scenario 2: Existing Project (No Docs)

Context: You have code but no formal documentation

Steps:

1. Install mini-coder-brain
2. Run /init-memory-bank (no arguments)
3. Claude auto-detects:
   • Tech stack from package.json
   • Project structure
   • Git history
   • Coding patterns
4. Confirm detected information
5. Provide missing details when prompted

Result: 75-85% quality (recommended)

What gets auto-detected:

• ✅ Project name (from package.json)
• ✅ Dependencies → tech stack
• ✅ File structure → frontend/backend
• ✅ Recent commits → progress
• ✅ Test files → testing framework

Scenario 3: Existing Project (With Docs)

Context: You have comprehensive documentation

Steps:

1. Install mini-coder-brain
2. Run /init-memory-bank --docs-path ./docs
3. Claude reads:
   • README.md
   • SRS.md
   • ARCHITECTURE.md
   • API.md
   • Any .md files in docs/
4. Extracts features, architecture, decisions
5. Generates rich context automatically

Result: 90-95% quality (optimal)

Documentation files recognized:

• SRS.md / REQUIREMENTS.md → Features
• ARCHITECTURE.md / DESIGN.md → System design
• API.md / API_SPEC.md → API patterns
• DECISIONS.md / ADR*.md → Technical decisions
• README.md → Project overview

🛠️ Essential Commands

/init-memory-bank

Purpose: Initialize project context (MANDATORY after installation)

Usage:

/init-memory-bank                      # Auto-detect
/init-memory-bank --docs-path ./docs   # With documentation
/init-memory-bank --dry-run            # Preview only

When to use: Once after installation, or after major project changes

/validate-context

Purpose: Check context quality with detailed report

Usage:

/validate-context                # Check quality
/validate-context --fix          # Interactive improvements

Output:

📊 Context Quality: 85% (Recommended) ✅

✅ Project name: defined
✅ Tech stack: 5 technologies
✅ Core features: 4 features
✅ Architecture: defined
⚠️  Development goals: missing

Recommendations:
  • Add development goals to productContext.md

When to use:

• After initialization
• When Claude seems confused
• Periodically to ensure quality

/update-memory-bank

Purpose: Update memory after major development work

Usage:

/update-memory-bank                              # Standard update
/update-memory-bank "Completed feature X"        # With note

When to use:

• After completing major features
• After making technical decisions
• End of development session
• When prompted by footer status

What gets updated:

• decisionLog.md - Technical decisions
• progress.md - Completed tasks
• activeContext.md - Current focus

/import-docs

Purpose: Import documentation after initialization

Usage:

/import-docs ./docs                     # Import folder
/import-docs ./specs/SRS.md             # Import file
/import-docs ./docs --type srs          # With type hint

When to use:

• Found documentation after initial setup
• Documentation was updated
• Want to add more context

/map-codebase

Purpose: Enable instant file access

Usage:

/map-codebase --rebuild       # First time (30 seconds)
/map-codebase                 # Load existing (instant)

When to use: Once per project, rebuild when structure changes

/memory-cleanup

Purpose: Archive old data to prevent “Prompt too long”

Usage:

/memory-cleanup                # Standard cleanup
/memory-cleanup --dry-run      # Preview only
/memory-cleanup --full         # Aggressive cleanup

When to use: When notified by status footer, or quarterly

🎓 Best Practices

1. Always Provide Documentation Paths

# ✅ GOOD: With documentation
/init-memory-bank --docs-path ./docs

# ❌ BAD: Without docs (when you have them)
/init-memory-bank

2. Validate Quality After Setup

/init-memory-bank --docs-path ./docs
/validate-context  # Always check!

Target: Aim for 80%+ quality

3. Update Memory After Major Work

# After completing features
/update-memory-bank "Completed authentication with JWT"

# After technical decisions
/update-memory-bank "Chose Redis for sessions"

4. Keep productContext.md Current

Manually update when:

• Adding major features
• Changing tech stack
• Updating project goals

5. Run Cleanup When Notified

When you see:

💡 Memory cleanup recommended: 12 session updates
   Run /memory-cleanup

Run: /memory-cleanup immediately

📊 Context Quality Guide

Quality Levels

Minimum Requirements

Must Have:

• ✅ Project name (not placeholder)
• ✅ Technology stack (3+ items)
• ✅ Core features (2+ items)

Should Have (for 80%+):

• ✅ Architecture description
• ✅ Development goals
• ✅ Project documentation

Improving Quality

From 60% → 80%:

1. Add architecture description
2. Define development goals
3. Document key patterns
4. Run /import-docs if you have documentation

From 80% → 95%:

1. Import comprehensive documentation
2. Add decision history
3. Document API patterns
4. Include success metrics

🚨 Troubleshooting

Problem: “Claude asks what framework I’m using”

Cause: Context not loaded or quality too low

Solution:

1. Run /validate-context
2. If quality <60%, run /validate-context --fix
3. Verify productContext.md has tech stack
4. Re-run /init-memory-bank if needed

Problem: “Context quality is 45%”

Cause: Insufficient information in memory bank

Solution:

1. Run /validate-context --fix
2. Follow interactive prompts
3. Or manually edit .claude/memory/productContext.md
4. Add: tech stack, features, architecture
5. Re-validate: /validate-context

Problem: “/init-memory-bank doesn’t populate files”

Cause: Templates exist but not initialized

Solution:

1. Check if files exist: ls .claude/memory/*.md
2. If they exist, back them up
3. Re-run installer
4. Run /init-memory-bank in Claude Code (not bash!)

Problem: “Prompt is too long” error

Cause: Memory bank bloat from many sessions

Solution:

1. Run /memory-cleanup immediately
2. Verify: check .claude/archive/ for backed up data
3. Quality should be maintained
4. Run cleanup quarterly

📚 Additional Resources

• SETUP.md: https://github.com/kunwar-shah/mini-coder-brain/blob/main/SETUP.md
• Examples: https://github.com/kunwar-shah/mini-coder-brain/tree/main/examples
• Commands: Full Command Reference
• Issues: https://github.com/kunwar-shah/mini-coder-brain/issues

✅ Quick Reference Card

# Installation
./install.sh /path/to/project

# Setup (MANDATORY)
/init-memory-bank --docs-path ./docs

# Validation
/validate-context

# Daily Use
/update-memory-bank "note"
/map-codebase

# Maintenance
/memory-cleanup  (when notified)

Ready to master Mini-CoderBrain? Follow these guides and reach 95% context quality!