MCP Infrastructure Architecture

Overview

The MCP (Model Context Protocol) infrastructure implements a distributed, event-driven architecture for managing MCP server instances. The system supports both containerized and URL-based MCP servers with dynamic configuration and secret management.

Architecture Patterns

1. Event-Driven Architecture

Pattern: Domain Events + Redis Pub/Sub Implementation:

Python FastAPI publishes domain events to Redis
Go MCP Manager subscribes to events and handles container lifecycle
Async communication ensures loose coupling

Python API ──(Domain Events)──> Redis ──(Pub/Sub)──> Go MCP Manager
     │                                                      │
     └── PostgreSQL Database                                └── Podman Containers

Key Components:

MCPServerInstanceCreated events trigger container creation
MCPServerInstanceDeleted events trigger container cleanup
Event payload includes complete configuration (json_spec)

2. Unified Configuration Pattern

Pattern: JSON Specification with Schema Validation Implementation: Industry-standard connector design

# MCPServer defines the schema
env_schema = [
    {
        "name": "API_KEY",
        "type": "string", 
        "description": "API key for service",
        "required": True,
        "secret": True
    }
]

# MCPServerInstance contains the actual configuration
json_spec = {
    "type": "docker",  # or "url"
    "image": "mcp/filesystem:latest",
    "port": 8001,
    "environment": {
        "API_KEY": "secret_ref:instance_123:API_KEY"
    },
    "command": ["python", "-m", "mcp_filesystem"],
    "resources": {
        "memory_limit": "256m",
        "cpu_limit": "0.5"
    }
}

3. Secret Management Pattern

Pattern: Reference-based Secret Storage Implementation:

Environment variables stored as references in json_spec
Actual values stored in secret manager (Infisical/DB)
Go service resolves secrets at container runtime

json_spec.environment.API_KEY = "secret_ref:instance_123:API_KEY"
                                      ↓
Secret Manager: "instance_123:API_KEY" → "actual_secret_value"
                                      ↓
Container Environment: API_KEY=actual_secret_value

4. Multi-Provider Pattern

Pattern: Type-based Provider Selection Implementation: Support for different MCP deployment types

Docker Provider

{
    "type": "docker",
    "image": "agentarea/echo:latest",
    "port": 8080,
    "environment": {...},
    "command": [...],
    "resources": {...}
}

URL Provider

{
    "type": "url",
    "endpoint": "http://localhost:3333/mcp",
    "health_check": "/health",
    "timeout": 30
}

5. Repository Pattern

Pattern: Domain-Driven Repository with Event Publishing Implementation:

Separate repositories for MCPServer and MCPServerInstance
Event publishing integrated into service layer
Clean separation of persistence and business logic

6. Dependency Injection Pattern

Pattern: FastAPI Dependency Injection Implementation:

Services injected via Depends()
Secret manager, event broker, repositories all injectable
Enables testing with mock implementations

Data Models

MCPServer (Schema Definition)

class MCPServer:
    id: UUID
    name: str
    description: str
    docker_image_url: str  # Default image for docker type
    version: str
    tags: List[str]
    status: str
    is_public: bool
    env_schema: List[Dict[str, Any]]  # Defines required environment variables

MCPServerInstance (Runtime Configuration)

class MCPServerInstance:
    id: UUID
    name: str
    description: Optional[str]
    server_spec_id: Optional[str]  # Reference to MCPServer
    json_spec: Dict[str, Any]      # Complete runtime configuration
    status: str

Environment Schema Structure

The env_schema in MCPServer defines the contract for environment variables:

env_schema = [
    {
        "name": "API_KEY",
        "type": "string",
        "description": "API key for external service",
        "required": True,
        "secret": True,
        "default": None
    },
    {
        "name": "MAX_CONNECTIONS", 
        "type": "integer",
        "description": "Maximum number of connections",
        "required": False,
        "secret": False,
        "default": "10"
    }
]

JSON Spec Structure

The json_spec in MCPServerInstance contains the complete runtime configuration:

Docker Type

{
    "type": "docker",
    "image": "mcp/filesystem:latest",
    "port": 8001,
    "environment": {
        "ALLOWED_DIRECTORIES": "/tmp,/var/tmp",
        "MAX_FILE_SIZE": "5MB",
        "API_KEY": "secret_ref:instance_123:API_KEY"
    },
    "command": ["python", "-m", "mcp_filesystem"],
    "resources": {
        "memory_limit": "256m", 
        "cpu_limit": "0.5"
    },
    "health_check": {
        "path": "/health",
        "interval": 30,
        "timeout": 10
    }
}

URL Type

{
    "type": "url",
    "endpoint": "http://localhost:3333/mcp",
    "health_check": {
        "path": "/health",
        "interval": 30,
        "timeout": 10
    },
    "authentication": {
        "type": "bearer",
        "token": "secret_ref:instance_123:AUTH_TOKEN"
    }
}

Service Architecture

Python FastAPI Services

MCPServerService: Manages server definitions and schemas
MCPServerInstanceService: Manages instance lifecycle and configuration
MCPEnvironmentService: Handles secret storage and retrieval
EventBroker: Publishes domain events to Redis

Go MCP Manager Services

EventSubscriber: Listens for Redis events
ContainerManager: Manages Podman containers
SecretResolver: Resolves secret references (to be implemented)
HealthChecker: Monitors container/URL health

Event Flow

Instance Creation:

API Request → MCPServerInstanceService → Database → Event Published → 
Redis → Go EventSubscriber → ContainerManager → Podman Container

Secret Resolution:

json_spec.environment → SecretResolver → Secret Manager → 
Actual Values → Container Environment

Health Monitoring:

HealthChecker → Container/URL Status → Database Update → 
Status Events → API Response

Network Architecture

┌─────────────────┐    ┌─────────────────┐    ┌─────────────────┐
│   Python API    │    │   Go Manager    │    │   MCP Containers│
│   (Port 8000)   │    │   (Port 7999)   │    │   (Dynamic)     │
└─────────────────┘    └─────────────────┘    └─────────────────┘
         │                       │                       │
         └───────────────────────┼───────────────────────┘
                                 │
         ┌─────────────────┐    ┌─────────────────┐
         │     Redis       │    │   PostgreSQL    │
         │   (Port 6379)   │    │   (Port 5432)   │
         └─────────────────┘    └─────────────────┘
                                 │
         ┌─────────────────┐    ┌─────────────────┐
         │     Traefik     │    │   Secret Store  │
         │   (Port 81)     │    │   (Infisical)   │
         └─────────────────┘    └─────────────────┘

Security Patterns

1. Secret Isolation

Secrets never stored in json_spec directly
Reference-based storage with resolution at runtime
Separate secret manager with encrypted storage

2. Network Isolation

MCP containers on isolated network
Reverse proxy (Traefik) for external access
Internal service communication only

3. Resource Limits

Container resource constraints (CPU, memory)
Maximum container limits per instance
Health check timeouts and retries

Implementation Status

✅ Completed

Event-driven architecture with Redis
Basic container management with Podman
Database models and repositories
API endpoints for CRUD operations
Network configuration and service discovery

🚧 In Progress

Event parsing and data extraction (minor bug fix needed)
Container creation from json_spec

📋 Future Enhancements

Secret resolution in Go service
URL-based MCP provider support
Health monitoring and status updates
Resource management and cleanup
Production deployment configuration

Testing Strategy

Integration Tests

End-to-end flow from API to container creation
Event publishing and consumption
Secret management integration
Health check validation

Test Containers

agentarea/echo: Simple HTTP echo server for testing
mcp/filesystem: Filesystem MCP server
URL endpoint: localhost:3333/mcp for URL-based testing

Configuration Examples

Example 1: Docker-based MCP Server

# Server definition
server = {
    "name": "filesystem-mcp",
    "description": "Filesystem access MCP server",
    "docker_image_url": "mcp/filesystem:latest",
    "env_schema": [
        {"name": "ALLOWED_DIRECTORIES", "type": "string", "required": True},
        {"name": "MAX_FILE_SIZE", "type": "string", "required": False, "default": "10MB"}
    ]
}

# Instance configuration
instance = {
    "name": "fs-instance-1",
    "server_spec_id": server_id,
    "json_spec": {
        "type": "docker",
        "image": "mcp/filesystem:latest",
        "port": 8001,
        "environment": {
            "ALLOWED_DIRECTORIES": "/tmp,/var/tmp",
            "MAX_FILE_SIZE": "5MB"
        }
    }
}

Example 2: URL-based MCP Server

# Server definition (minimal for URL type)
server = {
    "name": "external-mcp",
    "description": "External MCP service",
    "docker_image_url": "",  # Not used for URL type
    "env_schema": []
}

# Instance configuration
instance = {
    "name": "external-instance-1", 
    "server_spec_id": server_id,
    "json_spec": {
        "type": "url",
        "endpoint": "http://localhost:3333/mcp",
        "health_check": {
            "path": "/health",
            "interval": 30
        }
    }
}

This architecture provides a flexible, scalable foundation for managing MCP servers with strong separation of concerns, event-driven communication, and secure secret management.

​MCP Infrastructure Architecture

​Overview

​Architecture Patterns

​1. Event-Driven Architecture

​2. Unified Configuration Pattern

​3. Secret Management Pattern

​4. Multi-Provider Pattern

​Docker Provider

​URL Provider

​5. Repository Pattern

​6. Dependency Injection Pattern

​Data Models

​MCPServer (Schema Definition)

​MCPServerInstance (Runtime Configuration)

​Environment Schema Structure

​JSON Spec Structure

​Docker Type

​URL Type

​Service Architecture

​Python FastAPI Services

​Go MCP Manager Services

​Event Flow

​Network Architecture

​Security Patterns

​1. Secret Isolation

​2. Network Isolation

​3. Resource Limits

​Implementation Status

​✅ Completed

​🚧 In Progress

​📋 Future Enhancements

​Testing Strategy

​Integration Tests

​Test Containers

​Configuration Examples

​Example 1: Docker-based MCP Server

​Example 2: URL-based MCP Server