Skip to content

Latest commit

 

History

History
264 lines (189 loc) · 6.75 KB

README.md

File metadata and controls

264 lines (189 loc) · 6.75 KB

Memory MCP Server Claude Code Configuration 🧠

A production-grade Claude Code configuration specialized for building MCP servers with memory persistence, vector search, and AI companion systems.

✨ Features

This configuration provides comprehensive support for:

  • Memory Systems - Vector-indexed persistence with pgvector
  • MCP Protocol - Complete server implementation toolkit
  • Database Architecture - PostgreSQL 17 with Neon serverless
  • AI Companions - Multi-tenant architecture patterns
  • Production Deployment - Docker, Kubernetes, monitoring

📦 Installation

  1. Copy the .claude directory to your MCP server project:
cp -r memory-mcp-server/.claude your-mcp-project/
cp memory-mcp-server/CLAUDE.md your-mcp-project/
  1. The configuration will be automatically loaded when you start Claude Code.

🤖 Specialized Agents (15 total)

MCP Protocol Experts

Agent Description Use Cases
mcp-protocol-expert Protocol debugging and compliance Connection issues, protocol validation
mcp-sdk-builder SDK implementation patterns Building new MCP servers
mcp-transport-expert Transport layers (stdio, HTTP, SSE) Session management, optimization
mcp-types-expert TypeScript and Zod schemas Type safety, JSON-RPC formats

Database & Vector Search

Agent Description Use Cases
neon-drizzle-expert Neon PostgreSQL with Drizzle ORM Database setup, migrations
pgvector-advanced Advanced pgvector v0.8.0 features Binary vectors, HNSW indexes
vector-search-expert Semantic search and embeddings OpenAI embeddings, similarity search

Memory & Architecture

Agent Description Use Cases
memory-architecture Database design and indexing Schema design, retrieval optimization
memory-lifecycle Consolidation and expiration Memory decay models, deduplication
memory-validator Data integrity and validation CRUD operations, testing
companion-architecture Multi-tenant AI systems Isolation strategies, scaling

Development & Operations

Agent Description Use Cases
code-reviewer Comprehensive code review Security focus, best practices
debugger Systematic debugging Root cause analysis
test-runner Automated testing MCP protocol validation
production-deployment HTTPS deployment Containerization, monitoring

🛠️ Commands (7 total)

Development Workflow

/setup quick       # Quick project setup with essentials
/setup full        # Complete environment with all dependencies
/setup database    # Database-focused initialization

Testing & Review

/test             # Generate comprehensive test suites
/review           # Security-focused code review
/explain          # Context-aware code explanation

MCP Operations

/mcp-debug        # Debug MCP protocol issues
/memory-ops       # Test memory CRUD operations
/perf-monitor     # Performance profiling

🪝 Automation Hooks

TypeScript Development Hook

Automatically triggered on file modifications:

  • ✅ Type checking with tsc --noEmit
  • ✨ Prettier formatting
  • 🔧 ESLint fixing
  • 🧪 Test execution for test files
  • 📁 Smart filtering (skips node_modules, build dirs)

Command Logging

  • 📝 Logs all executed Bash commands
  • ⏱️ Timestamps for debugging
  • 📊 Audit trail maintenance

⚙️ Configuration Details

Security Permissions

{
  "permissions": {
    "allow": [
      "Read", "Grep", "Glob", "LS",
      "Bash(npm test:*)",
      "Write(**/*.ts)",
      "Bash(npx drizzle-kit:*)",
      "Bash(psql:*)"
    ],
    "deny": [
      "Read(./.env)",
      "Bash(rm -rf:*)",
      "Bash(git push:*)"
    ]
  }
}

Environment Variables

Pre-configured for MCP development:

  • DATABASE_URL - PostgreSQL connection
  • OPENAI_API_KEY - For embeddings
  • MCP_SERVER_PORT - Server configuration
  • NEON_DATABASE_URL - Serverless PostgreSQL

🚀 Usage Examples

Building an MCP Memory Server

# 1. Set up the project
> /setup full

# 2. Design memory schema
> Use memory-architecture agent to design the database schema

# 3. Implement MCP server
> Use mcp-sdk-builder agent to create the server

# 4. Add vector search
> Use vector-search-expert to implement semantic search

# 5. Deploy to production
> Use production-deployment agent for containerization

Debugging MCP Connections

# Debug protocol issues
> /mcp-debug

# The debugger will:
# - Validate protocol compliance
# - Check message formats
# - Test transport layer
# - Identify connection issues

📊 Technology Stack

Optimized for:

  • TypeScript & Node.js
  • PostgreSQL 17 with Neon serverless
  • Drizzle ORM v0.44.4 for type-safe database
  • pgvector v0.8.0 for vector similarity
  • @modelcontextprotocol/sdk for MCP
  • OpenAI embeddings for semantic search
  • Docker & Kubernetes for deployment

🎯 Key Features

Memory Persistence

  • Vector-indexed storage with pgvector
  • Semantic search capabilities
  • Memory consolidation and lifecycle
  • Multi-tenant isolation

MCP Protocol Support

  • Complete SDK implementation patterns
  • Transport layer optimization
  • Protocol compliance validation
  • Session management

Production Ready

  • Docker containerization
  • Kubernetes orchestration
  • Prometheus/Grafana monitoring
  • Structured logging

🔧 Customization

Edit .claude/settings.json to customize:

  • Permissions for your security needs
  • Environment variables for your services
  • Hook configurations for your workflow
  • Agent selections for your domain

📝 Best Practices

This configuration enforces:

  1. Type Safety - Full TypeScript with Zod validation
  2. Security First - Input validation, authentication
  3. Performance - Optimized vector search, caching
  4. Testing - Comprehensive test coverage
  5. Monitoring - Structured logging, metrics
  6. Documentation - Clear code comments, API docs

🐛 Troubleshooting

Common Issues

Hooks not executing:

chmod +x .claude/hooks/*.sh

Database connection issues:

# Check environment variables
echo $DATABASE_URL
# Test connection
psql $DATABASE_URL

MCP protocol errors:

/mcp-debug

📚 Resources

Built for production MCP server development 🚀

Transform your MCP server development with specialized AI assistance and automation.