smf-mcp 0.1.2

Enterprise-grade MCP framework built on FastMCP

SMF - Enterprise MCP Framework

SMF is a production-ready Python framework built on top of FastMCP that makes it significantly simpler and more professional to create, structure, and deploy MCP (Model Context Protocol) servers in enterprise production environments.

Features

• 🏗️ High-level abstractions: ServerFactory and AppBuilder for minimal boilerplate
• 🔧 Modular registration: Filesystem conventions, explicit registries, and plugin discovery
• ⚙️ Production runtime: Configuration system, structured logging, health endpoints, lifecycle management
• 🔐 Enterprise auth: Pluggable authentication (JWT, OAuth, OIDC) and authorization policies
• 🔄 Middleware pipeline: Chain of Responsibility pattern for logging, metrics, rate limiting, error handling
• 🔌 Plugin system: Extensible architecture with stable interfaces
• 📊 Observability: Prometheus metrics, OpenTelemetry tracing, structured logging
• 🚀 CLI & Templates: Project scaffolding and code generation

Architecture

┌─────────────────────────────┐
│   CLI & Templates           │
├─────────────────────────────┤
│ Integrations (AuthZ, AI SDK) │
├─────────────────────────────┤
│  Extensions & Middleware    │
├─────────────────────────────┤
│    Core SMF Layer          │
├─────────────────────────────┤
│    FastMCP (Upstream)        │
└─────────────────────────────┘

SMF wraps FastMCP without modifying it, ensuring compatibility with FastMCP updates while adding enterprise features.

Quick Start

Installation

uv add smf fastmcp

Simple Server

from smf import create_server

# Create server
mcp = create_server("My Server")

# Register a tool
@mcp.tool
def greet(name: str) -> str:
    """Greet someone by name."""
    return f"Hello, {name}!"

if __name__ == "__main__":
    mcp.run()

Advanced Server with AppBuilder

from smf import AppBuilder, Settings

# Custom settings
settings = Settings(
    server_name="My Server",
    structured_logging=True,
    metrics_enabled=True,
    rate_limit_enabled=True,
)

# Use AppBuilder
with AppBuilder(settings=settings) as builder:
    @builder.tool(tags=["math"])
    def add(a: float, b: float) -> float:
        """Add two numbers."""
        return a + b
    
    mcp = builder.build()

if __name__ == "__main__":
    mcp.run(transport="http", port=8000)

Configuration

SMF uses Pydantic-based settings that can be loaded from:

• Environment variables (prefix: SMF_)
• .env files
• YAML/JSON configuration files

# smf.yaml
server_name: "My Server"
server_version: "1.0.0"
transport: "http"
host: "0.0.0.0"
port: 8000
structured_logging: true
metrics_enabled: true
rate_limit_enabled: true
rate_limit_per_minute: 100
auth_provider: "jwt"
auth_config:
  secret: "${JWT_SECRET}"

CLI

# Initialize new project
smf init my-server

# Activate a plugin (e.g., Elasticsearch)
smf activate-plugin elasticsearch

# Add a tool
smf add-tool my-tool --description "Tool description"

# Validate configuration (syntax and values)
smf validate --config smf.yaml

# Validate with comprehensive tests
smf validate --config smf.yaml --tests

# Run server
smf run server.py --transport http --port 8000

# Inspect server (Official Web Inspector)
smf-inspector server.py

# Inspect server (Legacy Text Mode)
smf-inspector server.py --text

Core Components

ServerFactory

Creates configured FastMCP servers with defaults applied:

from smf import ServerFactory, Settings

factory = ServerFactory(settings=Settings())
mcp = factory.create(name="My Server")

AppBuilder

Fluent interface for registering components:

from smf import AppBuilder

builder = AppBuilder()
builder.tool(my_function)
builder.resource(my_resource)
mcp = builder.build()

Settings

Centralized configuration management:

from smf import Settings

settings = Settings(
    server_name="My Server",
    structured_logging=True,
    metrics_enabled=True,
)

Middleware

SMF provides built-in middleware:

• StructuredLoggingMiddleware: JSON-structured logging
• TracingMiddleware: OpenTelemetry distributed tracing
• RateLimitingMiddleware: Token bucket rate limiting
• ErrorHandlingMiddleware: Normalized error responses
• MetricsMiddleware: Prometheus metrics collection

Authentication

Pluggable authentication providers:

from smf import Settings, AuthProvider

settings = Settings(
    auth_provider=AuthProvider.JWT,
    auth_config={
        "secret": "your-secret-key",
        "algorithm": "HS256",
    }
)

Supported providers:

• JWT
• OAuth
• OIDC (via TokenVerifier)
• Custom providers

Plugins

Elasticsearch Plugin

Create Elasticsearch-powered MCP servers:

# Create server with CLI
smf init-elasticsearch my-server --index "products"

# Or use in code
from smf.plugins.elasticsearch import ElasticsearchClient, create_elasticsearch_tools

es_client = ElasticsearchClient(hosts="http://localhost:9200")
mcp = create_server("My Server")
tools = create_elasticsearch_tools(es_client, index="products")
for tool in tools:
    mcp.tool(tool)

Install: pip install smf[elasticsearch]

See Elasticsearch Plugin Documentation for details.

Examples

See src/examples/ for:

• simple_server/: Basic server setup
• advanced_server/: Advanced patterns with custom settings

Documentation

• FastMCP Research: FastMCP API mapping
• Architecture: Detailed architecture documentation

Requirements

• Python 3.11+
• FastMCP >= 2.11

Development

# Install dependencies
uv sync --dev

# Run tests
pytest

# Format code
black src/

# Type check
mypy src/

License

MIT

Credits

Built on FastMCP by Prefect.