Generic MCP (Model Context Protocol) server development patterns...
This skill provides comprehensive patterns and reusable code for building MCP (Model Context Protocol) servers that can expose any domain operations as tools for AI agents. Follows 2025 best practices for performance, security, and maintainability.
Use this skill when you need to:
# mcp_server/core.py
#!/usr/bin/env python3
"""
Generic MCP Server Base Architecture
Provides reusable patterns for any MCP server implementation
"""
import asyncio
import json
import logging
from abc import ABC, abstractmethod
from datetime import datetime, timedelta
from typing import Any, Dict, List, Optional, Sequence, Union, Callable
from contextlib import asynccontextmanager
from dataclasses import dataclass, field
from enum import Enum
import redis.asyncio as redis
from mcp.server import Server, NotificationOptions, stdio
from mcp.server.models import InitializationOptions
from mcp.server.stdio import stdio_server
from mcp.types import (
Resource, Tool, TextContent, ImageContent, EmbeddedResource,
LoggingLevel, CallToolRequest, EmptyResult,
ListResourcesRequest, ListToolsRequest, ReadResourceRequest,
GetPromptRequest, ListPromptsRequest
)
from pydantic import BaseModel, Field, validator
import aiofiles
import yaml
from pathlib import Path
# Configure logging
logging.basicConfig(
level=logging.INFO,
format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
)
logger = logging.getLogger("mcp_server")
class ServerConfig(BaseModel):
"""MCP Server configuration"""
name: str
version: str = "1.0.0"
description: str
debug: bool = False
redis_url: Optional[str] = None
rate_limit_requests: int = 100
rate_limit_window: int = 60
cache_ttl: int = 300
max_retries: int = 3
timeout: int = 30
class Config:
extra = "allow"
@dataclass
class RequestContext:
"""Request context for tool calls"""
user_id: str
session_id: Optional[str] = None
metadata: Dict[str, Any] = field(default_factory=dict)
timestamp: datetime = field(default_factory=datetime.utcnow)
class RateLimiter:
"""Redis-based rate limiter for MCP operations"""
def __init__(self, redis_url: str, requests: int, window: int):
self.redis_url = redis_url
self.requests = requests
self.window = window
self._redis = None
async def _get_redis(self):
if not self._redis:
self._redis = await redis.from_url(self.redis_url)
return self._redis
async def is_allowed(self, key: str) -> bool:
"""Check if request is allowed"""
r = await self._get_redis()
current = await r.incr(f"rate_limit:{key}")
if current == 1:
await r.expire(f"rate_limit:{key}", self.window)
return current <= self.requests
async def get_remaining(self, key: str) -> int:
"""Get remaining requests"""
r = await self._get_redis()
current = await r.get(f"rate_limit:{key}")
return max(0, self.requests - int(current or 0))
class CacheManager:
"""Redis-based caching for MCP responses"""
def __init__(self, redis_url: str, ttl: int = 300):
self.redis_url = redis_url
self.ttl = ttl
self._redis = None
async def _get_redis(self):
if not self._redis:
self._redis = await redis.from_url(self.redis_url)
return self._redis
def _make_key(self, tool_name: str, args: Dict[str, Any]) -> str:
"""Generate cache key from tool name and arguments"""
import hashlib
args_str = json.dumps(args, sort_keys=True)
return f"cache:{tool_name}:{hashlib.md5(args_str.encode()).hexdigest()}"
async def get(self, tool_name: str, args: Dict[str, Any]) -> Optional[Any]:
"""Get cached result"""
r = await self._get_redis()
key = self._make_key(tool_name, args)
result = await r.get(key)
return json.loads(result) if result else None
async def set(self, tool_name: str, args: Dict[str, Any], value: Any):
"""Cache result"""
r = await self._get_redis()
key = self._make_key(tool_name, args)
await r.setex(key, self.ttl, json.dumps(value))
class BaseMCPServer:
"""Base MCP Server with common functionality"""
def __init__(self, config: ServerConfig):
self.config = config
self.server = Server(config.name)
self.tools: Dict[str, Callable] = {}
self.rate_limiter: Optional[RateLimiter] = None
self.cache: Optional[CacheManager] = None
# Setup optional components
if config.redis_url:
self.rate_limiter = RateLimiter(
config.redis_url,
config.rate_limit_requests,
config.rate_limit_window
)
self.cache = CacheManager(
config.redis_url,
config.cache_ttl
)
# Register handlers
self._register_handlers()
logger.info(f"MCP Server '{config.name}' initialized")
def _register_handlers(self):
"""Register MCP handlers"""
@self.server.list_tools()
async def handle_list_tools() -> List[Tool]:
"""Return list of available tools"""
return await self.list_tools()
@self.server.call_tool()
async def handle_call_tool(name: str, arguments: Dict[str, Any]) -> List[TextContent]:
"""Handle tool call with rate limiting and caching"""
return await self.call_tool(name, arguments)
def register_tool(self, name: str, handler: Callable, schema: Dict[str, Any]):
"""Register a new tool"""
self.tools[name] = {
"handler": handler,
"schema": schema
}
logger.info(f"Registered tool: {name}")
async def list_tools(self) -> List[Tool]:
"""List all available tools"""
tools = []
for name, tool_info in self.tools.items():
tools.append(Tool(
name=name,
description=tool_info["schema"].get("description", ""),
inputSchema=tool_info["schema"].get("inputSchema", {})
))
return tools
async def call_tool(self, name: str, arguments: Dict[str, Any]) -> List[TextContent]:
"""Execute a tool call with full middleware pipeline"""
start_time = datetime.utcnow()
try:
# Extract context from arguments
context = self._extract_context(arguments)
# Rate limiting check
if self.rate_limiter:
rate_key = f"{context.user_id}:{name}"
if not await self.rate_limiter.is_allowed(rate_key):
return [TextContent(
type="text",
text=json.dumps({
"status": "error",
"error": "Rate limit exceeded",
"remaining": await self.rate_limiter.get_remaining(rate_key)
})
)]
# Check cache
if self.cache and self._is_cacheable(name):
cached_result = await self.cache.get(name, arguments)
if cached_result:
logger.info(f"Cache hit for tool: {name}")
return [TextContent(
type="text",
text=json.dumps(cached_result)
)]
# Validate tool exists
if name not in self.tools:
raise ValueError(f"Unknown tool: {name}")
# Validate arguments
schema = self.tools[name]["schema"]
self._validate_arguments(arguments, schema)
# Execute tool
handler = self.tools[name]["handler"]
result = await self._execute_tool(handler, arguments, context)
# Cache result if applicable
if self.cache and self._is_cacheable(name) and result.get("status") != "error":
await self.cache.set(name, arguments, result)
# Log execution
duration = (datetime.utcnow() - start_time).total_seconds()
logger.info(f"Tool {name} executed in {duration:.2f}s for user {context.user_id}")
return [TextContent(
type="text",
text=json.dumps(result, default=str)
)]
except Exception as e:
logger.error(f"Error executing tool {name}: {str(e)}", exc_info=True)
duration = (datetime.utcnow() - start_time).total_seconds()
logger.error(f"Tool {name} failed after {duration:.2f}s")
return [TextContent(
type="text",
text=json.dumps({
"status": "error",
"error": str(e),
"tool": name,
"timestamp": datetime.utcnow().isoformat()
})
)]
def _extract_context(self, arguments: Dict[str, Any]) -> RequestContext:
"""Extract request context from arguments"""
user_id = arguments.pop("_user_id", "anonymous")
session_id = arguments.pop("_session_id", None)
metadata = arguments.pop("_metadata", {})
return RequestContext(
user_id=user_id,
session_id=session_id,
metadata=metadata
)
def _validate_arguments(self, arguments: Dict[str, Any], schema: Dict[str, Any]):
"""Validate tool arguments against schema"""
# Basic validation - can be extended with pydantic
input_schema = schema.get("inputSchema", {})
required = input_schema.get("required", [])
properties = input_schema.get("properties", {})
# Check required fields
for field in required:
if field not in arguments:
raise ValueError(f"Missing required field: {field}")
# Validate field types
for field, value in arguments.items():
if field in properties:
field_schema = properties[field]
expected_type = field_schema.get("type")
if expected_type == "string" and not isinstance(value, str):
raise ValueError(f"Field {field} must be a string")
elif expected_type == "integer" and not isinstance(value, int):
raise ValueError(f"Field {field} must be an integer")
elif expected_type == "array" and not isinstance(value, list):
raise ValueError(f"Field {field} must be an array")
# Check enum values
if "enum" in field_schema and value not in field_schema["enum"]:
raise ValueError(f"Field {field} must be one of {field_schema['enum']}")
def _is_cacheable(self, tool_name: str) -> bool:
"""Determine if tool result should be cached"""
# Non-mutating operations are cacheable
non_mutating = ["get", "list", "search", "find", "read"]
return any(op in tool_name.lower() for op in non_mutating)
async def _execute_tool(self, handler: Callable, arguments: Dict[str, Any], context: RequestContext) -> Dict[str, Any]:
"""Execute tool handler with error handling"""
try:
# Pass context to handler if it accepts it
import inspect
sig = inspect.signature(handler)
if 'context' in sig.parameters:
result = await handler(arguments, context=context)
else:
result = await handler(arguments)
return result
except Exception as e:
logger.error(f"Tool handler failed: {str(e)}")
return {
"status": "error",
"error": str(e),
"timestamp": datetime.utcnow().isoformat()
}
async def run(self):
"""Start the MCP server"""
logger.info(f"Starting MCP server: {self.config.name}")
async with stdio_server() as (read_stream, write_stream):
await self.server.run(
read_stream,
write_stream,
InitializationOptions(
server_name=self.config.name,
server_version=self.config.version,
capabilities=self.server.get_capabilities(
notification_options=NotificationOptions(),
experimental_capabilities={},
)
)
)
def tool(
name: Optional[str] = None,
description: str = "",
input_schema: Optional[Dict[str, Any]] = None
):
"""Decorator for registering MCP tools"""
def decorator(func):
tool_name = name or func.__name__
schema = {
"description": description or func.__doc__ or "",
"inputSchema": input_schema or {}
}
# Store schema on function for later registration
func._mcp_tool_schema = schema
func._mcp_tool_name = tool_name
return func
return decorator
# mcp_server/database.py
"""
Generic Database Integration for MCP Servers
Supports multiple ORMs and connection patterns
"""
import asyncio
from abc import ABC, abstractmethod
from contextlib import asynccontextmanager
from typing import Any, Dict, List, Optional, TypeVar, Generic, Union
from datetime import datetime
import json
from sqlalchemy import create_engine, MetaData, Table, Column, Integer, String, DateTime, Text, Boolean, select, update, delete, insert
from sqlalchemy.ext.asyncio import create_async_engine, AsyncSession, async_sessionmaker
from sqlalchemy.orm import sessionmaker, declarative_base
from sqlalchemy.pool import NullPool
import asyncpg
import motor.motor_asyncio
from redis.asyncio import Redis
# Type variables
T = TypeVar('T')
class DatabaseBackend(ABC):
"""Abstract base for database backends"""
@abstractmethod
async def connect(self):
"""Establish connection"""
pass
@abstractmethod
async def disconnect(self):
"""Close connection"""
pass
@abstractmethod
async def execute_query(self, query: str, params: Dict[str, Any] = None) -> List[Dict[str, Any]]:
"""Execute a query and return results"""
pass
@abstractmethod
async def execute_command(self, command: str, params: Dict[str, Any] = None) -> Any:
"""Execute a command (INSERT, UPDATE, DELETE)"""
pass
class PostgresBackend(DatabaseBackend):
"""PostgreSQL backend using asyncpg"""
def __init__(self, connection_string: str):
self.connection_string = connection_string
self.pool: Optional[asyncpg.Pool] = None
async def connect(self):
self.pool = await asyncpg.create_pool(
self.connection_string,
min_size=5,
max_size=20,
command_timeout=60
)
async def disconnect(self):
if self.pool:
await self.pool.close()
async def execute_query(self, query: str, params: Dict[str, Any] = None) -> List[Dict[str, Any]]:
async with self.pool.acquire() as conn:
rows = await conn.fetch(query, *params.values() if params else [])
return [dict(row) for row in rows]
async def execute_command(self, command: str, params: Dict[str, Any] = None) -> Any:
async with self.pool.acquire() as conn:
return await conn.execute(command, *params.values() if params else [])
class SQLAlchemyBackend(DatabaseBackend):
"""SQLAlchemy backend for multiple databases"""
def __init__(self, database_url: str, async_mode: bool = True):
self.database_url = database_url
self.async_mode = async_mode
self.engine = None
self.session_factory = None
async def connect(self):
if self.async_mode:
self.engine = create_async_engine(
self.database_url,
pool_pre_ping=True,
pool_recycle=300,
echo=False
)
self.session_factory = async_sessionmaker(
self.engine,
class_=AsyncSession,
expire_on_commit=False
)
else:
self.engine = create_engine(
self.database_url,
pool_pre_ping=True,
pool_recycle=300,
echo=False
)
self.session_factory = sessionmaker(
bind=self.engine,
expire_on_commit=False
)
async def disconnect(self):
if self.engine:
await self.engine.dispose()
@asynccontextmanager
async def get_session(self):
"""Get database session"""
async with self.session_factory() as session:
try:
yield session
if self.async_mode:
await session.commit()
else:
session.commit()
except Exception:
if self.async_mode:
await session.rollback()
else:
session.rollback()
raise
finally:
if self.async_mode:
await session.close()
else:
session.close()
async def execute_query(self, query: Any, params: Dict[str, Any] = None) -> List[Dict[str, Any]]:
"""Execute SQLAlchemy query"""
async with self.get_session() as session:
if isinstance(query, str):
# Raw SQL query
result = await session.execute(query, params or {})
rows = result.fetchall()
return [dict(row._mapping) for row in rows]
else:
# SQLAlchemy ORM query
result = await session.execute(query)
rows = result.fetchall()
return [dict(row._mapping) for row in rows]
async def execute_command(self, command: Any, params: Dict[str, Any] = None) -> Any:
"""Execute SQLAlchemy command"""
async with self.get_session() as session:
if isinstance(command, str):
# Raw SQL command
result = await session.execute(command, params or {})
await session.commit()
return result
else:
# SQLAlchemy ORM command
await session.execute(command, params or {})
await session.commit()
return None
class MongoBackend(DatabaseBackend):
"""MongoDB backend using motor"""
def __init__(self, connection_string: str, database_name: str):
self.connection_string = connection_string
self.database_name = database_name
self.client = None
self.db = None
async def connect(self):
self.client = motor.motor_asyncio.AsyncIOMotorClient(self.connection_string)
self.db = self.client[self.database_name]
async def disconnect(self):
if self.client:
self.client.close()
async def execute_query(self, collection: str, query: Dict[str, Any] = None) -> List[Dict[str, Any]]:
"""Execute MongoDB find query"""
cursor = self.db[collection].find(query or {})
results = []
async for document in cursor:
# Convert ObjectId to string
if '_id' in document:
document['_id'] = str(document['_id'])
results.append(document)
return results
async def execute_command(self, operation: str, collection: str, data: Dict[str, Any]) -> Any:
"""Execute MongoDB command"""
if operation == "insert":
result = await self.db[collection].insert_one(data)
return str(result.inserted_id)
elif operation == "update":
filter_ = data.pop("_filter")
update_data = {"$set": data}
result = await self.db[collection].update_one(filter_, update_data)
return result.modified_count
elif operation == "delete":
result = await self.db[collection].delete_one(data)
return result.deleted_count
class DatabaseManager(Generic[T]):
"""Generic database manager for MCP servers"""
def __init__(self, backend: DatabaseBackend):
self.backend = backend
self._connected = False
async def connect(self):
"""Connect to database"""
if not self._connected:
await self.backend.connect()
self._connected = True
async def disconnect(self):
"""Disconnect from database"""
if self._connected:
await self.backend.disconnect()
self._connected = False
@asynccontextmanager
async def transaction(self):
"""Database transaction context manager"""
if hasattr(self.backend, 'get_session'):
async with self.backend.get_session() as session:
yield session
else:
# For backends that don't support transactions
yield self.backend
async def find_one(self, table_or_collection: str, query: Dict[str, Any]) -> Optional[Dict[str, Any]]:
"""Find a single record"""
if isinstance(self.backend, MongoBackend):
results = await self.backend.execute_query(table_or_collection, query)
return results[0] if results else None
else:
# SQL implementation
where_clause = " AND ".join([f"{k} = :{k}" for k in query.keys()])
sql = f"SELECT * FROM {table_or_collection} WHERE {where_clause} LIMIT 1"
results = await self.backend.execute_query(sql, query)
return results[0] if results else None
async def find_many(
self,
table_or_collection: str,
query: Dict[str, Any] = None,
limit: int = None,
offset: int = None,
order_by: str = None
) -> List[Dict[str, Any]]:
"""Find multiple records"""
query = query or {}
if isinstance(self.backend, MongoBackend):
cursor = self.backend.db[table_or_collection].find(query)
if limit:
cursor = cursor.limit(limit)
if offset:
cursor = cursor.skip(offset)
if order_by:
# MongoDB sort format
sort_field, sort_dir = order_by.split()
cursor = cursor.sort([(sort_field, 1 if sort_dir == 'ASC' else -1)])
results = []
async for document in cursor:
if '_id' in document:
document['_id'] = str(document['_id'])
results.append(document)
return results
else:
# SQL implementation
where_clause = ""
if query:
where_clause = "WHERE " + " AND ".join([f"{k} = :{k}" for k in query.keys()])
sql = f"SELECT * FROM {table_or_collection} {where_clause}"
if order_by:
sql += f" ORDER BY {order_by}"
if limit:
sql += f" LIMIT {limit}"
if offset:
sql += f" OFFSET {offset}"
return await self.backend.execute_query(sql, query)
async def create(self, table_or_collection: str, data: Dict[str, Any]) -> Any:
"""Create a new record"""
data = data.copy()
# Add timestamps
data['created_at'] = datetime.utcnow()
data['updated_at'] = datetime.utcnow()
if isinstance(self.backend, MongoBackend):
return await self.backend.execute_command("insert", table_or_collection, data)
else:
# SQL implementation
columns = list(data.keys())
placeholders = [f":{col}" for col in columns]
sql = f"INSERT INTO {table_or_collection} ({', '.join(columns)}) VALUES ({', '.join(placeholders)})"
return await self.backend.execute_command(sql, data)
async def update(self, table_or_collection: str, query: Dict[str, Any], data: Dict[str, Any]) -> int:
"""Update records"""
data = data.copy()
data['updated_at'] = datetime.utcnow()
if isinstance(self.backend, MongoBackend):
data['_filter'] = query
return await self.backend.execute_command("update", table_or_collection, data)
else:
# SQL implementation
where_clause = " AND ".join([f"{k} = :{k}" for k in query.keys()])
set_clause = ", ".join([f"{k} = :update_{k}" for k in data.keys()])
# Prefix update params to avoid conflicts
update_params = {f"update_{k}": v for k, v in data.items()}
params = {**query, **update_params}
sql = f"UPDATE {table_or_collection} SET {set_clause} WHERE {where_clause}"
result = await self.backend.execute_command(sql, params)
return result.rowcount if hasattr(result, 'rowcount') else 0
async def delete(self, table_or_collection: str, query: Dict[str, Any]) -> int:
"""Delete records"""
if isinstance(self.backend, MongoBackend):
return await self.backend.execute_command("delete", table_or_collection, query)
else:
# SQL implementation
where_clause = " AND ".join([f"{k} = :{k}" for k in query.keys()])
sql = f"DELETE FROM {table_or_collection} WHERE {where_clause}"
result = await self.backend.execute_command(sql, query)
return result.rowcount if hasattr(result, 'rowcount') else 0
async def count(self, table_or_collection: str, query: Dict[str, Any] = None) -> int:
"""Count records"""
query = query or {}
if isinstance(self.backend, MongoBackend):
return await self.backend.db[table_or_collection].count_documents(query)
else:
# SQL implementation
where_clause = ""
if query:
where_clause = "WHERE " + " AND ".join([f"{k} = :{k}" for k in query.keys()])
sql = f"SELECT COUNT(*) as count FROM {table_or_collection} {where_clause}"
results = await self.backend.execute_query(sql, query)
return results[0]['count'] if results else 0
# Factory function for creating database managers
def create_database_manager(database_url: str, backend_type: str = "auto") -> DatabaseManager:
"""Create database manager based on URL or backend type"""
if backend_type == "auto":
if database_url.startswith("postgresql+asyncpg://"):
backend = SQLAlchemyBackend(database_url, async_mode=True)
elif database_url.startswith("mongodb://"):
import re
match = re.match(r'mongodb://[^/]+/([^?]*)', database_url)
db_name = match.group(1) if match else "default"
backend = MongoBackend(database_url, db_name)
elif database_url.startswith("postgresql://"):
backend = PostgresBackend(database_url)
else:
backend = SQLAlchemyBackend(database_url, async_mode=True)
else:
if backend_type == "postgres":
backend = PostgresBackend(database_url)
elif backend_type == "mongodb":
db_name = database_url.split("/")[-1].split("?")[0]
backend = MongoBackend(database_url, db_name)
elif backend_type == "sqlalchemy":
backend = SQLAlchemyBackend(database_url)
else:
raise ValueError(f"Unknown backend type: {backend_type}")
return DatabaseManager(backend)
# mcp_server/tools.py
"""
Generic MCP Tool Implementation Patterns
"""
from typing import Any, Dict, List, Optional, Union, Callable
from datetime import datetime, timedelta
import json
import uuid
from dataclasses import dataclass, field
from .core import BaseMCPServer, tool, RequestContext
from .database import DatabaseManager
class BaseTool:
"""Base class for MCP tools"""
def __init__(self, db_manager: DatabaseManager, cache=None):
self.db_manager = db_manager
self.cache = cache
async def execute(self, args: Dict[str, Any], context: RequestContext = None) -> Dict[str, Any]:
"""Execute the tool logic"""
raise NotImplementedError
def _validate_permissions(self, context: RequestContext, required_permission: str = None) -> bool:
"""Validate user permissions"""
# Implement permission checking logic
return True
class CRUDBaseTool(BaseTool):
"""Base CRUD tool for any entity"""
def __init__(self, table_name: str, db_manager: DatabaseManager, schema: Dict[str, Any]):
super().__init__(db_manager)
self.table_name = table_name
self.schema = schema
self.entity_name = table_name.rstrip('s') # Remove plural 's'
async def create(self, args: Dict[str, Any], context: RequestContext) -> Dict[str, Any]:
"""Create entity"""
try:
# Validate against schema
validated_data = self._validate_data(args, for_create=True)
# Add user context
if context:
validated_data['user_id'] = context.user_id
if context.session_id:
validated_data['session_id'] = context.session_id
# Insert into database
result = await self.db_manager.create(self.table_name, validated_data)
return {
"status": "created",
"id": result,
"entity": self.entity_name,
"timestamp": datetime.utcnow().isoformat()
}
except Exception as e:
return {
"status": "error",
"error": str(e),
"entity": self.entity_name,
"operation": "create"
}
async def get(self, args: Dict[str, Any], context: RequestContext) -> Dict[str, Any]:
"""Get entity by ID"""
try:
entity_id = args.get("id")
if not entity_id:
raise ValueError("Missing required field: id")
# Add user filter for security
query = {"id": entity_id}
if context and not self._validate_permissions(context, "read_all"):
query["user_id"] = context.user_id
result = await self.db_manager.find_one(self.table_name, query)
if not result:
return {
"status": "not_found",
"entity": self.entity_name,
"id": entity_id
}
return {
"status": "success",
"entity": self.entity_name,
"data": self._serialize_data(result)
}
except Exception as e:
return {
"status": "error",
"error": str(e),
"entity": self.entity_name,
"operation": "get"
}
async def list(self, args: Dict[str, Any], context: RequestContext) -> Dict[str, Any]:
"""List entities with filtering"""
try:
# Build query from args
query = {}
filters = args.get("filters", {})
limit = args.get("limit", 20)
offset = args.get("offset", 0)
order_by = args.get("order_by", "created_at DESC")
# Add user filter for security
if context and not self._validate_permissions(context, "read_all"):
query["user_id"] = context.user_id
# Apply additional filters
query.update(filters)
# Fetch from database
results = await self.db_manager.find_many(
self.table_name,
query=query,
limit=limit,
offset=offset,
order_by=order_by
)
# Get total count
total = await self.db_manager.count(self.table_name, query)
return {
"status": "success",
"entity": self.entity_name,
"data": [self._serialize_data(r) for r in results],
"pagination": {
"total": total,
"limit": limit,
"offset": offset,
"has_more": offset + limit < total
}
}
except Exception as e:
return {
"status": "error",
"error": str(e),
"entity": self.entity_name,
"operation": "list"
}
async def update(self, args: Dict[str, Any], context: RequestContext) -> Dict[str, Any]:
"""Update entity"""
try:
entity_id = args.pop("id", None)
if not entity_id:
raise ValueError("Missing required field: id")
# Validate update data
update_data = self._validate_data(args, for_create=False)
# Build query filter
query = {"id": entity_id}
if context and not self._validate_permissions(context, "update_all"):
query["user_id"] = context.user_id
# Update in database
affected = await self.db_manager.update(self.table_name, query, update_data)
if affected == 0:
return {
"status": "not_found",
"entity": self.entity_name,
"id": entity_id
}
return {
"status": "updated",
"entity": self.entity_name,
"id": entity_id,
"affected_rows": affected,
"timestamp": datetime.utcnow().isoformat()
}
except Exception as e:
return {
"status": "error",
"error": str(e),
"entity": self.entity_name,
"operation": "update"
}
async def delete(self, args: Dict[str, Any], context: RequestContext) -> Dict[str, Any]:
"""Delete entity"""
try:
entity_id = args.get("id")
if not entity_id:
raise ValueError("Missing required field: id")
# Build query filter
query = {"id": entity_id}
if context and not self._validate_permissions(context, "delete_all"):
query["user_id"] = context.user_id
# Delete from database
affected = await self.db_manager.delete(self.table_name, query)
if affected == 0:
return {
"status": "not_found",
"entity": self.entity_name,
"id": entity_id
}
return {
"status": "deleted",
"entity": self.entity_name,
"id": entity_id,
"affected_rows": affected,
"timestamp": datetime.utcnow().isoformat()
}
except Exception as e:
return {
"status": "error",
"error": str(e),
"entity": self.entity_name,
"operation": "delete"
}
def _validate_data(self, data: Dict[str, Any], for_create: bool = False) -> Dict[str, Any]:
"""Validate data against schema"""
validated = {}
schema_fields = self.schema.get("properties", {})
required_fields = self.schema.get("required", [])
# Check required fields for create
if for_create:
for field in required_fields:
if field not in data:
raise ValueError(f"Missing required field: {field}")
# Validate each field
for field, value in data.items():
if field not in schema_fields:
continue # Skip unknown fields or raise error based on strictness
field_schema = schema_fields[field]
field_type = field_schema.get("type")
# Type validation
if field_type == "string":
if not isinstance(value, str):
raise ValueError(f"Field {field} must be a string")
# Check min/max length
if "minLength" in field_schema and len(value) < field_schema["minLength"]:
raise ValueError(f"Field {field} is too short")
if "maxLength" in field_schema and len(value) > field_schema["maxLength"]:
raise ValueError(f"Field {field} is too long")
elif field_type == "integer":
if not isinstance(value, int):
raise ValueError(f"Field {field} must be an integer")
# Check min/max value
if "minimum" in field_schema and value < field_schema["minimum"]:
raise ValueError(f"Field {field} is too small")
if "maximum" in field_schema and value > field_schema["maximum"]:
raise ValueError(f"Field {field} is too large")
elif field_type == "array":
if not isinstance(value, list):
raise ValueError(f"Field {field} must be an array")
# Check enum values
if "enum" in field_schema and value not in field_schema["enum"]:
raise ValueError(f"Field {field} must be one of {field_schema['enum']}")
validated[field] = value
return validated
def _serialize_data(self, data: Dict[str, Any]) -> Dict[str, Any]:
"""Serialize data for output"""
serialized = data.copy()
# Handle datetime serialization
for key, value in serialized.items():
if isinstance(value, datetime):
serialized[key] = value.isoformat()
elif isinstance(value, dict):
# Convert complex types to JSON string
try:
json.dumps(value)
except TypeError:
serialized[key] = str(value)
return serialized
class BulkOperationTool(BaseTool):
"""Tool for bulk operations on entities"""
def __init__(self, table_name: str, db_manager: DatabaseManager, schema: Dict[str, Any]):
super().__init__(db_manager)
self.table_name = table_name
self.schema = schema
self.entity_name = table_name.rstrip('s')
async def bulk_create(self, args: Dict[str, Any], context: RequestContext) -> Dict[str, Any]:
"""Bulk create entities"""
try:
items = args.get("items", [])
if not items:
raise ValueError("No items provided for bulk create")
# Validate all items
validated_items = []
for item in items:
validated = self._validate_item(item)
if context:
validated["user_id"] = context.user_id
validated_items.append(validated)
# Insert all items
results = []
for item in validated_items:
result = await self.db_manager.create(self.table_name, item)
results.append(result)
return {
"status": "created",
"entity": self.entity_name,
"count": len(results),
"ids": results,
"timestamp": datetime.utcnow().isoformat()
}
except Exception as e:
return {
"status": "error",
"error": str(e),
"entity": self.entity_name,
"operation": "bulk_create"
}
async def bulk_update(self, args: Dict[str, Any], context: RequestContext) -> Dict[str, Any]:
"""Bulk update entities"""
try:
updates = args.get("updates", [])
if not updates:
raise ValueError("No updates provided")
total_affected = 0
for update in updates:
entity_id = update.get("id")
update_data = update.get("data", {})
if not entity_id:
continue
# Build query
query = {"id": entity_id}
if context:
query["user_id"] = context.user_id
# Update
affected = await self.db_manager.update(self.table_name, query, update_data)
total_affected += affected
return {
"status": "updated",
"entity": self.entity_name,
"affected_rows": total_affected,
"updates_processed": len(updates),
"timestamp": datetime.utcnow().isoformat()
}
except Exception as e:
return {
"status": "error",
"error": str(e),
"entity": self.entity_name,
"operation": "bulk_update"
}
async def bulk_delete(self, args: Dict[str, Any], context: RequestContext) -> Dict[str, Any]:
"""Bulk delete entities"""
try:
ids = args.get("ids", [])
if not ids:
raise ValueError("No IDs provided for bulk delete")
total_affected = 0
for entity_id in ids:
# Build query
query = {"id": entity_id}
if context:
query["user_id"] = context.user_id
# Delete
affected = await self.db_manager.delete(self.table_name, query)
total_affected += affected
return {
"status": "deleted",
"entity": self.entity_name,
"affected_rows": total_affected,
"ids_processed": len(ids),
"timestamp": datetime.utcnow().isoformat()
}
except Exception as e:
return {
"status": "error",
"error": str(e),
"entity": self.entity_name,
"operation": "bulk_delete"
}
def _validate_item(self, item: Dict[str, Any]) -> Dict[str, Any]:
"""Validate a single item"""
# Use CRUD base tool validation
crud_tool = CRUDBaseTool(self.table_name, self.db_manager, self.schema)
return crud_tool._validate_data(item, for_create=True)
# examples/task_mcp_server.py
"""
Example: Task Management MCP Server using generic patterns
"""
import os
from typing import Dict, Any
from mcp_server.core import BaseMCPServer, ServerConfig, tool
from mcp_server.database import create_database_manager
from mcp_server.tools import CRUDBaseTool, BulkOperationTool
# Server configuration
config = ServerConfig(
name="task-manager",
version="1.0.0",
description="Generic task management MCP server",
redis_url=os.getenv("REDIS_URL", "redis://localhost:6379"),
database_url=os.getenv("DATABASE_URL", "postgresql+asyncpg://user:pass@localhost/tasks")
)
# Task entity schema
TASK_SCHEMA = {
"type": "object",
"properties": {
"title": {
"type": "string",
"minLength": 1,
"maxLength": 200,
"description": "Task title"
},
"description": {
"type": "string",
"maxLength": 1000,
"description": "Task description"
},
"priority": {
"type": "string",
"enum": ["low", "medium", "high"],
"default": "medium",
"description": "Task priority"
},
"status": {
"type": "string",
"enum": ["todo", "in_progress", "completed"],
"default": "todo",
"description": "Task status"
},
"due_date": {
"type": "string",
"format": "date-time",
"description": "Optional due date"
},
"tags": {
"type": "array",
"items": {"type": "string"},
"description": "Task tags"
}
},
"required": ["title"]
}
class TaskMCPServer(BaseMCPServer):
"""Task Management MCP Server"""
def __init__(self, config: ServerConfig):
super().__init__(config)
# Initialize database
self.db_manager = create_database_manager(config.database_url)
# Initialize tools
self.task_tool = CRUDBaseTool("tasks", self.db_manager, TASK_SCHEMA)
self.bulk_tool = BulkOperationTool("tasks", self.db_manager, TASK_SCHEMA)
# Register tools
self._register_task_tools()
def _register_task_tools(self):
"""Register all task-related tools"""
# Create task
self.register_tool(
"create_task",
self.task_tool.create,
{
"description": "Create a new task",
"inputSchema": {
"type": "object",
"properties": {
"title": {"type": "string", "description": "Task title"},
"description": {"type": "string", "description": "Optional description"},
"priority": {"type": "string", "enum": ["low", "medium", "high"]},
"due_date": {"type": "string", "format": "date-time"},
"tags": {"type": "array", "items": {"type": "string"}}
},
"required": ["title"]
}
}
)
# Get task
self.register_tool(
"get_task",
self.task_tool.get,
{
"description": "Get a task by ID",
"inputSchema": {
"type": "object",
"properties": {
"id": {"type": "integer", "description": "Task ID"}
},
"required": ["id"]
}
}
)
# List tasks
self.register_tool(
"list_tasks",
self.task_tool.list,
{
"description": "List tasks with optional filtering",
"inputSchema": {
"type": "object",
"properties": {
"filters": {"type": "object", "description": "Filter criteria"},
"limit": {"type": "integer", "minimum": 1, "maximum": 100, "default": 20},
"offset": {"type": "integer", "minimum": 0, "default": 0},
"order_by": {"type": "string", "description": "Order by field (e.g., 'created_at DESC')"}
}
}
}
)
# Update task
self.register_tool(
"update_task",
self.task_tool.update,
{
"description": "Update a task",
"inputSchema": {
"type": "object",
"properties": {
"id": {"type": "integer", "description": "Task ID"},
"title": {"type": "string", "description": "New title"},
"description": {"type": "string", "description": "New description"},
"priority": {"type": "string", "enum": ["low", "medium", "high"]},
"status": {"type": "string", "enum": ["todo", "in_progress", "completed"]},
"due_date": {"type": "string", "format": "date-time"},
"tags": {"type": "array", "items": {"type": "string"}}
},
"required": ["id"]
}
}
)
# Delete task
self.register_tool(
"delete_task",
self.task_tool.delete,
{
"description": "Delete a task",
"inputSchema": {
"type": "object",
"properties": {
"id": {"type": "integer", "description": "Task ID"}
},
"required": ["id"]
}
}
)
# Bulk create
self.register_tool(
"bulk_create_tasks",
self.bulk_tool.bulk_create,
{
"description": "Create multiple tasks at once",
"inputSchema": {
"type": "object",
"properties": {
"items": {
"type": "array",
"items": {
"type": "object",
"properties": {
"title": {"type": "string"},
"description": {"type": "string"},
"priority": {"type": "string", "enum": ["low", "medium", "high"]},
"tags": {"type": "array", "items": {"type": "string"}}
},
"required": ["title"]
}
}
},
"required": ["items"]
}
}
)
# Search tasks
self.register_tool(
"search_tasks",
self._search_tasks,
{
"description": "Search tasks by text query",
"inputSchema": {
"type": "object",
"properties": {
"query": {"type": "string", "description": "Search query"},
"limit": {"type": "integer", "minimum": 1, "maximum": 50, "default": 20}
},
"required": ["query"]
}
}
)
async def _search_tasks(self, args: Dict[str, Any], context) -> Dict[str, Any]:
"""Search tasks by text"""
try:
query = args.get("query", "")
limit = args.get("limit", 20)
# Build search query
if isinstance(self.db_manager.backend, MongoBackend):
# MongoDB text search
search_query = {
"$text": {"$search": query}
}
if context and not self.task_tool._validate_permissions(context, "read_all"):
search_query["user_id"] = context.user_id
results = await self.db_manager.find_many("tasks", search_query, limit=limit)
else:
# PostgreSQL full-text search
sql = """
SELECT * FROM tasks
WHERE to_tsvector('english', title || ' ' || COALESCE(description, '')) @@ plainto_tsquery('english', :query)
"""
params = {"query": query}
if context and not self.task_tool._validate_permissions(context, "read_all"):
sql += " AND user_id = :user_id"
params["user_id"] = context.user_id
sql += f" LIMIT {limit}"
results = await self.db_manager.execute_query(sql, params)
return {
"status": "success",
"entity": "task",
"data": [self.task_tool._serialize_data(r) for r in results],
"query": query,
"count": len(results)
}
except Exception as e:
return {
"status": "error",
"error": str(e),
"operation": "search_tasks"
}
# Main execution
async def main():
"""Start the Task MCP Server"""
server = TaskMCPServer(config)
await server.run()
if __name__ == "__main__":
import asyncio
asyncio.run(main())
# tests/test_mcp_server.py
"""
Generic MCP Server Testing Patterns
"""
import pytest
import asyncio
from typing import Dict, Any, List
from unittest.mock import Mock, AsyncMock
from mcp_server.core import BaseMCPServer, ServerConfig, RequestContext
from mcp_server.database import DatabaseManager, create_database_manager
from mcp_server.tools import CRUDBaseTool
class MockDatabaseManager:
"""Mock database manager for testing"""
def __init__(self):
self.data = {}
self.next_id = 1
async def create(self, table: str, data: Dict[str, Any]) -> int:
"""Mock create"""
entity_id = self.next_id
data['id'] = entity_id
self.data[f"{table}:{entity_id}"] = data
self.next_id += 1
return entity_id
async def find_one(self, table: str, query: Dict[str, Any]) -> Dict[str, Any]:
"""Mock find one"""
for key, value in self.data.items():
table_name, entity_id = key.split(":")
if table_name == table:
match = True
for k, v in query.items():
if value.get(k) != v:
match = False
break
if match:
return value
return None
async def find_many(self, table: str, query: Dict[str, Any], limit: int = None) -> List[Dict[str, Any]]:
"""Mock find many"""
results = []
for key, value in self.data.items():
table_name, entity_id = key.split(":")
if table_name == table:
match = True
for k, v in query.items():
if value.get(k) != v:
match = False
break
if match:
results.append(value)
if limit and len(results) >= limit:
break
return results
async def update(self, table: str, query: Dict[str, Any], data: Dict[str, Any]) -> int:
"""Mock update"""
count = 0
for key, value in self.data.items():
table_name, entity_id = key.split(":")
if table_name == table:
match = True
for k, v in query.items():
if value.get(k) != v:
match = False
break
if match:
value.update(data)
count += 1
return count
async def delete(self, table: str, query: Dict[str, Any]) -> int:
"""Mock delete"""
to_delete = []
for key, value in self.data.items():
table_name, entity_id = key.split(":")
if table_name == table:
match = True
for k, v in query.items():
if value.get(k) != v:
match = False
break
if match:
to_delete.append(key)
for key in to_delete:
del self.data[key]
return len(to_delete)
@pytest.fixture
async def mock_db():
"""Mock database manager fixture"""
return MockDatabaseManager()
@pytest.fixture
def test_schema():
"""Test entity schema"""
return {
"type": "object",
"properties": {
"name": {"type": "string", "minLength": 1},
"value": {"type": "integer", "minimum": 0},
"status": {"type": "string", "enum": ["active", "inactive"]},
"tags": {"type": "array", "items": {"type": "string"}}
},
"required": ["name"]
}
@pytest.fixture
def crud_tool(mock_db, test_schema):
"""CRUD tool fixture"""
return CRUDBaseTool("test_entities", mock_db, test_schema)
@pytest.fixture
def user_context():
"""User context fixture"""
return RequestContext(user_id="test_user", session_id="test_session")
class TestCRUDTool:
"""Test CRUD tool operations"""
@pytest.mark.asyncio
async def test_create_entity(self, crud_tool, user_context):
"""Test entity creation"""
args = {
"name": "Test Entity",
"value": 100,
"status": "active",
"tags": ["test", "example"]
}
result = await crud_tool.create(args, user_context)
assert result["status"] == "created"
assert "id" in result
assert result["entity"] == "test_entity"
assert "timestamp" in result
@pytest.mark.asyncio
async def test_create_missing_required(self, crud_tool, user_context):
"""Test creation with missing required field"""
args = {"value": 100} # Missing 'name'
result = await crud_tool.create(args, user_context)
assert result["status"] == "error"
assert "Missing required field" in result["error"]
@pytest.mark.asyncio
async def test_get_entity(self, crud_tool, user_context):
"""Test getting an entity"""
# First create an entity
create_args = {"name": "Test Get"}
create_result = await crud_tool.create(create_args, user_context)
entity_id = create_result["id"]
# Get the entity
get_args = {"id": entity_id}
result = await crud_tool.get(get_args, user_context)
assert result["status"] == "success"
assert result["data"]["name"] == "Test Get"
assert result["data"]["id"] == entity_id
@pytest.mark.asyncio
async def test_get_not_found(self, crud_tool, user_context):
"""Test getting non-existent entity"""
args = {"id": 99999}
result = await crud_tool.get(args, user_context)
assert result["status"] == "not_found"
@pytest.mark.asyncio
async def test_list_entities(self, crud_tool, user_context):
"""Test listing entities"""
# Create a few entities
for i in range(3):
args = {"name": f"Entity {i}"}
await crud_tool.create(args, user_context)
# List entities
result = await crud_tool.list({}, user_context)
assert result["status"] == "success"
assert len(result["data"]) == 3
assert "pagination" in result
assert result["pagination"]["total"] == 3
@pytest.mark.asyncio
async def test_list_with_filters(self, crud_tool, user_context):
"""Test listing with filters"""
# Create entities with different statuses
await crud_tool.create({"name": "Active 1", "status": "active"}, user_context)
await crud_tool.create({"name": "Inactive 1", "status": "inactive"}, user_context)
await crud_tool.create({"name": "Active 2", "status": "active"}, user_context)
# Filter by status
result = await crud_tool.list(
{"filters": {"status": "active"}},
user_context
)
assert result["status"] == "success"
assert all(entity["status"] == "active" for entity in result["data"])
@pytest.mark.asyncio
async def test_update_entity(self, crud_tool, user_context):
"""Test updating an entity"""
# Create entity
create_result = await crud_tool.create({"name": "Original"}, user_context)
entity_id = create_result["id"]
# Update entity
update_args = {
"id": entity_id,
"name": "Updated",
"value": 200
}
result = await crud_tool.update(update_args, user_context)
assert result["status"] == "updated"
assert result["affected_rows"] == 1
@pytest.mark.asyncio
async def test_delete_entity(self, crud_tool, user_context):
"""Test deleting an entity"""
# Create entity
create_result = await crud_tool.create({"name": "To Delete"}, user_context)
entity_id = create_result["id"]
# Delete entity
delete_args = {"id": entity_id}
result = await crud_tool.delete(delete_args, user_context)
assert result["status"] == "deleted"
assert result["affected_rows"] == 1
# Verify deletion
get_result = await crud_tool.get({"id": entity_id}, user_context)
assert get_result["status"] == "not_found"
class TestMCPServer:
"""Test MCP Server functionality"""
@pytest.mark.asyncio
async def test_server_initialization(self):
"""Test server initialization"""
config = ServerConfig(
name="test-server",
description="Test MCP Server"
)
server = BaseMCPServer(config)
assert server.config.name == "test-server"
assert server.server.name == "test-server"
assert len(server.tools) == 0
@pytest.mark.asyncio
async def test_tool_registration(self):
"""Test tool registration"""
config = ServerConfig(name="test-server", description="Test")
server = BaseMCPServer(config)
# Register a test tool
async def test_tool(args: Dict[str, Any]) -> Dict[str, Any]:
return {"status": "success", "data": args}
schema = {
"description": "Test tool",
"inputSchema": {
"type": "object",
"properties": {
"message": {"type": "string"}
}
}
}
server.register_tool("test_tool", test_tool, schema)
assert "test_tool" in server.tools
assert server.tools["test_tool"]["schema"] == schema
@pytest.mark.asyncio
async def test_rate_limiting(self):
"""Test rate limiting functionality"""
# This would require mocking Redis or using a test instance
# Implementation would depend on your rate limiting strategy
pass
@pytest.mark.asyncio
async def test_caching(self):
"""Test caching functionality"""
# This would require mocking Redis or using a test instance
# Implementation would depend on your caching strategy
pass
# Integration test example
class TestMCPServerIntegration:
"""Integration tests for MCP Server"""
@pytest.mark.asyncio
async def test_full_crud_workflow(self):
"""Test full CRUD workflow through MCP interface"""
# Create mock server
config = ServerConfig(
name="test-integration",
description="Integration test server"
)
# Use in-memory database
mock_db = MockDatabaseManager()
# Create CRUD tool
test_schema = {
"type": "object",
"properties": {
"name": {"type": "string"},
"value": {"type": "integer"}
},
"required": ["name"]
}
crud_tool = CRUDBaseTool("items", mock_db, test_schema)
# Register tools
server = BaseMCPServer(config)
server.register_tool(
"create_item",
crud_tool.create,
{
"description": "Create item",
"inputSchema": {
"type": "object",
"properties": {
"name": {"type": "string"},
"value": {"type": "integer"}
},
"required": ["name"]
}
}
)
# Test create
result = await server.call_tool(
"create_item",
{"name": "Test Item", "value": 123}
)
response = json.loads(result[0].text)
assert response["status"] == "created"
# Test get (would need get_item tool)
# Test list
# Test update
# Test delete
This generic MCP server skill provides a reusable foundation for building any MCP server, not just for todos. It includes:
★ Insight ─────────────────────────────────────
The key architectural pattern here is the separation of concerns between:
This makes the system highly reusable and maintainable. Any developer can quickly build a new MCP server by defining their entity schema and combining the generic tools.
─────────────────────────────────────────────────
