MCP Configuration Guide: YAML Setup and Configuration

What is MCP?

The Modular Copilot Platform (MCP) is a framework for deploying and managing AI assistants with modular capabilities. It allows you to create custom AI workflows with specialized tools and integrations.

Basic YAML Configuration

Minimal Configuration

# mcp-config.yml
version: "1.0"

server:
  host: "0.0.0.0"
  port: 5000

models:
  - name: "default"
    provider: "ollama"
    model: "mistral:7b"
    temperature: 0.7

tools:
  - filesystem
  - web_search

Configuration Breakdown

Server Settings:

server:
  host: "0.0.0.0"        # Listen on all interfaces
  port: 5000             # API port
  workers: 4             # Number of worker processes
  timeout: 120           # Request timeout (seconds)

Model Configuration:

models:
  - name: "code-assistant"
    provider: "ollama"    # Options: ollama, openai, anthropic
    model: "codellama:7b"
    temperature: 0.3      # Lower = more deterministic
    max_tokens: 2048
    context_window: 4096

Tools and Capabilities:

tools:
  - name: "filesystem"
    enabled: true
    config:
      allowed_paths:
        - "/home/user/documents"
        - "/home/user/projects"
      
  - name: "web_search"
    enabled: true
    api_key: "${SEARCH_API_KEY}"
    
  - name: "code_execution"
    enabled: false        # Disabled for security

Docker Deployment

docker-compose.yml

version: '3.8'

services:
  mcp:
    image: modularcopilot/mcp:latest
    container_name: mcp-server
    restart: unless-stopped
    ports:
      - "5000:5000"
    volumes:
      - ./config:/app/config
      - ./data:/app/data
    environment:
      - MCP_CONFIG_PATH=/app/config/mcp-config.yml
      - OLLAMA_HOST=http://ollama:11434
    depends_on:
      - ollama

  ollama:
    image: ollama/ollama:latest
    container_name: ollama
    restart: unless-stopped
    ports:
      - "11434:11434"
    volumes:
      - ollama_data:/root/.ollama

volumes:
  ollama_data:

Deployment Commands

# Start services
docker-compose up -d

# View logs
docker-compose logs -f mcp

# Test API
curl http://localhost:5000/health

# Pull model
docker exec ollama ollama pull mistral:7b

Advanced Configuration

Multi-Model Setup

models:
  - name: "general"
    provider: "ollama"
    model: "mistral:7b"
    use_case: "general_queries"
    
  - name: "coding"
    provider: "ollama"
    model: "codellama:13b"
    use_case: "code_generation"
    
  - name: "creative"
    provider: "openai"
    model: "gpt-4"
    api_key: "${OPENAI_API_KEY}"
    use_case: "creative_writing"

Custom Tool Configuration

tools:
  - name: "obsidian_integration"
    type: "custom"
    script: "/app/tools/obsidian_tool.py"
    config:
      vault_path: "/data/obsidian"
      api_endpoint: "http://localhost:27124"
      
  - name: "n8n_workflows"
    type: "webhook"
    url: "http://n8n:5678/webhook/mcp"
    headers:
      Authorization: "Bearer ${N8N_TOKEN}"

Environment Variables

# .env file
MCP_LOG_LEVEL=INFO
OLLAMA_HOST=http://localhost:11434
OPENAI_API_KEY=sk-...
SEARCH_API_KEY=...
N8N_TOKEN=...

Integration Examples

With n8n

// n8n HTTP Request Node
{
  "method": "POST",
  "url": "http://mcp:5000/api/chat",
  "body": {
    "model": "general",
    "messages": [
      {
        "role": "user",
        "content": "Summarize my Obsidian notes"
      }
    ]
  }
}

With Obsidian

// Obsidian plugin integration
async function queryMCP(prompt: string) {
  const response = await fetch('http://localhost:5000/api/chat', {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      model: 'general',
      messages: [{ role: 'user', content: prompt }]
    })
  });
  
  return await response.json();
}

Security Best Practices

1. Authentication

server:
  auth:
    enabled: true
    type: "bearer"
    tokens:
      - "${MCP_API_TOKEN}"

2. Rate Limiting

server:
  rate_limit:
    enabled: true
    requests_per_minute: 60
    burst: 10

3. Tool Restrictions

tools:
  - name: "filesystem"
    enabled: true
    security:
      read_only: true
      allowed_extensions: [".md", ".txt"]
      max_file_size: "10MB"

Troubleshooting

Common Issues

MCP won’t start:

# Check logs
docker logs mcp-server

# Validate config
docker exec mcp-server python -c "import yaml; yaml.safe_load(open('/app/config/mcp-config.yml'))"

Model not loading:

# Pull model manually
docker exec ollama ollama pull mistral:7b

# Check available models
docker exec ollama ollama list

Connection refused:

# Check if services are running
docker ps

# Test connectivity
curl http://localhost:5000/health
curl http://localhost:11434/api/version

Conclusion

MCP provides a flexible framework for AI integration. Proper YAML configuration is key to reliable deployment. Start with basic setup, then gradually add tools and models as needed.

Key Takeaways: - Use environment variables for secrets - Start with minimal configuration - Test each tool before enabling - Monitor resource usage - Keep configurations in version control

For more information: MCP GitHub

Mcp Configuration Guide Yaml Setup