### Example Claude Desktop Configuration with Multiple MCP Servers

Source: https://www.claudemcp.com/docs/en/quickstart

An example of a `claude_desktop_config.json` file demonstrating the integration of both the filesystem and sqlite MCP servers. This allows Claude Desktop to manage local files and interact with SQLite databases.

```json
{
  "mcpServers": {
    "sqlite": {
      "command": "uvx",
      "args": ["mcp-server-sqlite", "--db-path", "/Users/cnych/test.db"]
    },
    "filesystem": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "/Users/cnych/src"
      ]
    }
  }
}
```

--------------------------------

### Install Dependencies using Homebrew (macOS)

Source: https://www.claudemcp.com/docs/en/quickstart

Installs necessary tools like uv, git, and sqlite3 on macOS using the Homebrew package manager. These are prerequisites for setting up the MCP server.

```bash
# Using Homebrew
brew install uv git sqlite3
```

--------------------------------

### Example MCP Interactive Client Session

Source: https://www.claudemcp.com/docs/en/streamable-http

This is a sample output from running the MCP interactive client. It demonstrates the connection process, the list of available commands, and example interactions such as calling the 'greet' and 'multi-greet' tools, and starting a notification stream. The output also shows tool results and notifications received from the server.

```bash
$ node build/client/simpleStreamableHttp.js
MCP Interactive Client
=====================
Connecting to http://localhost:3000/mcp...
Transport created with session ID: a0fdb476-a82f-47bb-927f-58fcc332b5ea
Connected to MCP server

Available commands:
  connect [url]              - Connect to MCP server (default: http://localhost:3000/mcp)
  disconnect                 - Disconnect from server
  terminate-session          - Terminate the current session
  reconnect                  - Reconnect to the server
  list-tools                 - List available tools
  call-tool <name> [args]    - Call a tool with optional JSON arguments
  greet [name]               - Call the greet tool
  multi-greet [name]         - Call the multi-greet tool with notifications
  start-notifications [interval] [count] - Start periodic notifications
  list-prompts               - List available prompts
  get-prompt [name] [args]   - Get a prompt with optional JSON arguments
  list-resources             - List available resources
  help                       - Show this help
  quit                       - Exit the program

> list-tools
Available tools:
  - greet: A simple greeting tool
  - multi-greet: A tool that sends different greetings with delays between them
  - start-notification-stream: Starts sending periodic notifications for testing resumability

> greet james
Calling tool 'greet' with args: { name: 'james' }
Tool result:
  Hello, james!

> multi-greet claudemcp.com
Calling multi-greet tool with notifications...
Calling tool 'multi-greet' with args: { name: 'claudemcp.com' }

Notification #1: debug - Starting multiple greetings for claudemcp.com
> 
Notification #2: info - Sending first greeting to claudemcp.com
> 
Notification #3: info - Sending second greeting to claudemcp.com
> Tool result:
  Good morning, claudemcp.com!

> start-notifications 1000 5
Starting notification stream: interval=1000ms, count=5
Calling tool 'start-notification-stream' with args: { interval: 1000, count: 5 }

Notification #4: info - Periodic notification #1 at 2025-04-24T11:30:54.650Z
> 
Notification #5: info - Periodic notification #2 at 2025-04-24T11:30:55.652Z
> 
Notification #6: info - Periodic notification #3 at 2025-04-24T11:30:56.655Z
> 
Notification #7: info - Periodic notification #4 at 2025-04-24T11:30:57.656Z
> 
Notification #8: info - Periodic notification #5 at 2025-04-24T11:30:58.658Z
> Tool result:
  Started sending periodic notifications every 1000ms

>
```

--------------------------------

### Install Context7 MCP via Smithery CLI

Source: https://www.claudemcp.com/servers/en/context7

This command installs the Context7 MCP server for Claude Desktop using the Smithery CLI. It's a straightforward way to get the server set up for automatic integration.

```bash
npx -y @smithery/cli install @upstash/context7-mcp --client claude
```

--------------------------------

### Example Prompts for iOS Simulator Interaction

Source: https://www.claudemcp.com/servers/ko/ios-simulator-mcp

A collection of example prompts demonstrating various interactions with the iOS Simulator MCP Server. These prompts cover inspecting accessibility elements, inputting text, simulating taps and swipes, getting UI element details, taking screenshots, and recording/stopping video.

```text
현재 화면의 모든 접근성 요소 확인
텍스트 입력 필드에 "QA 테스트"를 입력하고 입력이 올바른지 확인
좌표 x=250, y=400을 탭하고 예상 요소가 트리거되는지 확인
x=150, y=600에서 x=150, y=100으로 스와이프하고 올바른 동작 확인
위치 x=300, y=350의 UI 요소를 설명하여 적절한 라벨링과 기능 확인
현재 시뮬레이터 화면의 스크린샷을 촬영하여 my_screenshot.png로 저장
시뮬레이터 화면 녹화 시작 (기본적으로 ~/Downloads/simulator_recording_$DATE.mp4에 저장)
현재 시뮬레이터 화면 녹화 중지
```

--------------------------------

### LangGraph Agent Client Example (Python)

Source: https://www.claudemcp.com/blogs/tw/langchain-mcp-adapters

This Python snippet demonstrates the client-side setup for a LangGraph agent that utilizes MCP tools. It shows how to initialize the agent and connect to the MCP server to leverage its tools within the LangGraph framework. Further details on agent implementation would follow.

```python
# This is a placeholder for the client-side code.
# The actual implementation would involve importing necessary libraries
# from langchain and langchain_mcp_adapters to define the agent and its tools.

# Example structure (not fully functional without context):
# from langchain_mcp_adapters import MCPClient
# from langgraph.graph import StateGraph, END
# from langchain_openai import ChatOpenAI
# from langchain.agents import AgentExecutor, create_tool_calling_agent, load_tools

# # Initialize MCP client
# mcp_client = MCPClient(server_url="http://localhost:8000") # Assuming a running MCP server

# # Load tools from MCP server
# mcp_tools = mcp_client.get_tools()

# # Combine with other tools if needed
# all_tools = load_tools([], llm=ChatOpenAI())
# all_tools.extend(mcp_tools)

# # Define the agent using LangGraph
# workflow = StateGraph("YourAgentState") # Replace YourAgentState with the actual state type
# # ... add nodes and edges for the agent workflow ...
# app = workflow.compile()

# # Run the agent
# result = app.invoke({"input": "What is 2 + 2?"})
# print(result)
```

--------------------------------

### Manual Installation Method 2: uvx for Direct Installation

Source: https://www.claudemcp.com/servers/tw/dify-mcp-server

This JSON configuration demonstrates how to install and run the Dify MCP Server directly from its Git repository using 'uvx' without needing to clone the code locally. It specifies the command, the Git URL, and the target Python module to run, along with the configuration file environment variable. This is the recommended manual installation method.

```json
"mcpServers": {
  "dify-mcp-server": {
    "command": "uvx",
      "args": [
        "--from","git+https://github.com/YanxingLiu/dify-mcp-server","dify_mcp_server"
      ],
    "env": {
       "CONFIG_PATH": "/Users/lyx/Downloads/config.yaml"
    }
  }
}
```

--------------------------------

### Install and Configure Supabase MCP Server (JSON & Shell)

Source: https://www.claudemcp.com/servers/tw/supabase-mcp

This snippet shows how to configure the Supabase MCP server client, including specifying the command to run and arguments like the access token. It also provides a command-line example for running the server with read-only mode.

```json
{
  "mcpServers": {
    "supabase": {
      "command": "npx",
      "args": [
        "-y",
        "@supabase/mcp-server-supabase@latest",
        "--access-token",
        "<personal-access-token>"
      ]
    }
  }
}
```

```json
{
  "mcpServers": {
    "supabase": {
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "@supabase/mcp-server-supabase@latest",
        "--access-token",
        "<personal-access-token>"
      ]
    }
  }
}
```

```shell
npx -y @supabase/mcp-server-supabase@latest --access-token=<personal-access-token> --read-only
```

--------------------------------

### Install Dependencies and Playwright

Source: https://www.claudemcp.com/servers/en/browser-use-mcp-server

Installs project dependencies using uv, installs the Playwright Python package, and downloads the necessary browser binaries for Playwright automation.

```bash
uv sync
uv pip install playwright
uv run playwright install --with-deps --no-shell chromium
```

--------------------------------

### Development Setup Commands (Bash)

Source: https://www.claudemcp.com/servers/zh/firecrawl-mcp

Standard commands for setting up and running the development environment for the project. This includes installing dependencies, building the project, and running tests. These commands are in Bash.

```bash
# 安装依赖
npm install

# 构建
npm run build

# 运行测试
npm test
```

--------------------------------

### Install LangChain MCP Adapter

Source: https://www.claudemcp.com/blogs/en/langchain-mcp-adapters

Instructions for installing the LangChain MCP Adapter using pip and uv. Dependencies like langgraph and langchain-openai are also mentioned for uv installation.

```bash
pip install langchain-mcp-adapters
```

```bash
uv add langchain-mcp-adapters langgraph langchain-openai
```

--------------------------------

### Install Dependencies for Context7 MCP Development

Source: https://www.claudemcp.com/servers/en/context7

This command installs the necessary dependencies for developing the Context7 MCP project using Bun. It's the initial step after cloning the repository.

```bash
bun i
```

--------------------------------

### Run mcp-server-fetch as a script

Source: https://www.claudemcp.com/servers/en/fetch

This snippet demonstrates how to run the mcp-server-fetch server as a Python script after installation via pip. This command starts the server process.

```bash
python -m mcp_server_fetch
```

--------------------------------

### Build and Install Local Package

Source: https://www.claudemcp.com/servers/en/browser-use-mcp-server

Commands to build a distributable wheel for the local package and install it globally. This is part of the local development workflow to test changes.

```bash
# From the project root directory
uv build
uv tool uninstall browser-use-mcp-server 2>/dev/null || true
uv tool install dist/browser_use_mcp_server-*.whl
```

--------------------------------

### Install Development Dependencies for SEO MCP

Source: https://www.claudemcp.com/servers/ko/seo-mcp

This command installs the necessary dependencies for developing the SEO MCP project after cloning the repository. It uses 'uv sync' for efficient dependency management, which is faster than pip install.

```bash
uv sync  # or pip install -e .
```

--------------------------------

### Local Development Setup (Bash)

Source: https://www.claudemcp.com/servers/tw/claudflare

Instructions for setting up the project for local development. This involves installing project dependencies using pnpm, building the project with watch mode enabled, and then running the local server for testing and integration with MCP clients.

```bash
pnpm install
pnpm build:watch

# Then, in a second terminal:
node dist/index.js init
```

--------------------------------

### Install iOS Simulator MCP Server Locally

Source: https://www.claudemcp.com/servers/ko/ios-simulator-mcp

This snippet outlines the steps for setting up the iOS Simulator MCP Server from its source code. It includes cloning the repository, installing dependencies with npm, building the project, and configuring Cursor MCP to use the local build.

```bash
git clone https://github.com/joshuayoes/ios-simulator-mcp
cd ios-simulator-mcp
npm install
npm run build
```

```json
{
  "mcpServers": {
    "ios-simulator": {
      "command": "node",
      "args": ["/path/to/your/ios-simulator-mcp/build/index.js"]
    }
  }
}
```

--------------------------------

### Install mcp-server-fetch using PIP

Source: https://www.claudemcp.com/servers/en/fetch

This snippet shows how to install the mcp-server-fetch package using pip. Ensure Node.js is installed if you need the more robust HTML simplifier.

```bash
pip install mcp-server-fetch
```

--------------------------------

### Client Execution Example

Source: https://www.claudemcp.com/docs/en/write-ts-client

Example commands demonstrating how to build and run the MCP client. It shows the typical command-line usage with a path to the server script and the expected output during a test interaction.

```bash
node build/index.js /Users/cnych/src/weather-server/build/index.js
# output
Weather MCP server running on stdio
Connected to server with tools: [ 'get_forecast' ]

MCP Client Started!
Type your queries or 'exit' to exit.

Query: Beijing weather for the next 3 days

I can get the weather forecast for Beijing for the next 3 days. Let me do that for you now.
[Calling tool get_forecast with args {"city":"Beijing","days":3}]
# Beijing Weather Forecast

## Next 3 Days:
- **March 24, 2025**: 21°C with overcast clouds
- **March 25, 2025**: 21°C with scattered clouds
- **March 26, 2025**: 20°C with overcast clouds

The temperature will remain relatively stable around 20-21°C over the next three days, with cloudy conditions throughout. March 25th will have partially clear skies with scattered clouds, while the other days will be predominantly overcast.

Query:
```

--------------------------------

### Install SEO MCP Dependencies for Development

Source: https://www.claudemcp.com/servers/tw/seo-mcp

These commands clone the SEO MCP repository and install its development dependencies using 'uv sync'. This is an alternative to 'pip install -e .' for setting up the development environment.

```bash
git clone https://github.com/cnych/seo-mcp.git
cd seo-mcp
uv sync
```

--------------------------------

### Example uv Configuration for Dify MCP Server

Source: https://www.claudemcp.com/servers/en/dify-mcp-server

An example JSON configuration for running the Dify MCP Server using 'uv'. This configuration includes specific paths for the server directory and the configuration file, demonstrating how to set up the environment for local execution.

```json
{
  "mcpServers": {
    "dify-mcp-server": {
      "command": "uv",
      "args": [
        "--directory",
        "/Users/lyx/Downloads/dify-mcp-server",
        "run",
        "dify_mcp_server"
      ],
      "env": {
        "CONFIG_PATH": "/Users/lyx/Downloads/config.yaml"
      }
    }
  }
}
```

--------------------------------

### Install SEO MCP using pip

Source: https://www.claudemcp.com/servers/en/seo-mcp

Installs the SEO MCP package from PyPI using pip. Ensure you have Python 3.10 or higher and pip installed.

```bash
pip install seo-mcp
```

--------------------------------

### Install Local Development Build

Source: https://www.claudemcp.com/servers/ko/browser-use-mcp-server

Uninstalls any existing global installation of the server and then installs the newly built wheel file. This is for testing local changes.

```bash
# Install as a global tool
uv tool uninstall browser-use-mcp-server 2>/dev/null || true
uv tool install dist/browser_use_mcp_server-*.whl
```

--------------------------------

### Install SEO MCP using uv

Source: https://www.claudemcp.com/servers/en/seo-mcp

Installs the SEO MCP package using the `uv` package manager. This is an alternative to pip and requires `uv` to be installed.

```bash
uv pip install seo-mcp
```

--------------------------------

### Install Prerequisites with uv

Source: https://www.claudemcp.com/servers/en/browser-use-mcp-server

Installs necessary prerequisites like uv and mcp-proxy, which are essential for running the browser-use-mcp-server. uv is a fast Python package manager, and mcp-proxy is required for stdio mode.

```bash
curl -LsSf https://astral.sh/uv/install.sh | sh
uv tool install mcp-proxy
uv tool update-shell
```

--------------------------------

### Install and Run Server in stdio Mode

Source: https://www.claudemcp.com/servers/en/browser-use-mcp-server

Builds the project into a wheel, installs it globally as a tool, and then runs the server in stdio mode. This mode is used for communication over standard input/output, often integrated with other tools.

```bash
# 1. Build and install globally
uv build
uv tool uninstall browser-use-mcp-server 2>/dev/null || true
uv tool install dist/browser_use_mcp_server-*.whl

# 2. Run with stdio transport
browser-use-mcp-server run server --port 8000 --stdio --proxy-port 9000
```

--------------------------------

### Install LangChain MCP Adapters with Uv

Source: https://www.claudemcp.com/blogs/tw/langchain-mcp-adapters

This command installs the LangChain MCP Adapters library along with other necessary packages (langgraph, langchain-openai) using the uv package manager. This is an alternative installation method for users of uv.

```bash
uv add langchain-mcp-adapters langgraph langchain-openai
```

--------------------------------

### Build and Install for stdio Mode

Source: https://www.claudemcp.com/servers/ko/browser-use-mcp-server

Builds a distributable wheel for the project and installs it as a global tool using uv. This prepares the server for operation in stdio (standard input/output) mode.

```bash
# 1. Build and globally install
uv build
uv tool uninstall browser-use-mcp-server 2>/dev/null || true
uv tool install dist/browser_use_mcp_server-*.whl
```

--------------------------------

### Install Dependencies with npm

Source: https://www.claudemcp.com/servers/en/web-search

Installs the necessary Node.js dependencies for the web search server project using npm.

```bash
npm install

```

--------------------------------

### Start Figma Context MCP Server

Source: https://www.claudemcp.com/servers/ko/figma-context-mcp

Command to start the Figma Context MCP server using npx. Requires replacing '<your-figma-api-key>' with the actual Figma access token.

```bash
npx figma-developer-mcp --figma-api-key=<your-figma-api-key>
```

--------------------------------

### Example Prompts with Context7 MCP

Source: https://www.claudemcp.com/servers/en/context7

These examples demonstrate how to use the 'use context7' command within prompts for LLMs, enabling them to fetch up-to-date information. The prompts are designed to elicit specific code generation tasks.

```txt
Create a basic Next.js project with app router. use context7
```

```txt
Create a script to delete the rows where the city is "" given PostgreSQL credentials. use context7
```

--------------------------------

### Tool: webcrawl_sites example

Source: https://www.claudemcp.com/servers/zh/mcp-server-webcrawl

An example of how to use the `webcrawl_sites` tool to list all crawled sites.

```bash
# 你能列出网络爬虫吗？
```

--------------------------------

### Programmatic Server Implementation

Source: https://www.claudemcp.com/servers/en/playwright

Guide on how to integrate Playwright MCP directly into a Node.js application for custom server setup and transport mechanisms.

```APIDOC
## Programmatic Server Implementation

### Description
This section outlines how to programmatically create and manage a Playwright MCP server within a Node.js application, offering greater control over server options and transport layers.

### Method
Node.js Script

### Endpoint
N/A

### Parameters
#### Request Body (for `createServer`)
- **launchOptions** (object) - Optional - Playwright launch options (e.g., `{ headless: true }`).
- **Other server options** (if available) - Specific to the MCP server implementation.

### Request Example
```javascript
import { createServer } from "@playwright/mcp";
// Import necessary transport classes if needed, e.g., from '@playwright/mcp/lib/sseServerTransport';

async function runMyMCPServer() {
  const server = createServer({
    launchOptions: {
      headless: true,
      // other Playwright options...
    },
    // Specify other server options if available
  });

  // Connect to transports (e.g., SSE, custom)
  // Example for SSE transport (conceptual, requires HTTP server setup):
  /*
  const http = require('http');
  const { SSEServerTransport } = require('@playwright/mcp/lib/sseServerTransport'); // Adjust path as needed

  const httpServer = http.createServer((req, res) => {
    if (req.url === '/messages' && req.method === 'GET') {
      res.writeHead(200, {
        'Content-Type': 'text/event-stream',
        'Cache-Control': 'no-cache',
        'Connection': 'keep-alive',
      });
      const transport = new SSEServerTransport("/messages", res);
      server.connect(transport);

      req.on('close', () => {
        server.disconnect(transport);
      });
    } else {
      res.writeHead(404);
      res.end();
    }
  });

  httpServer.listen(8931, () => {
    console.log('MCP Server with SSE transport listening on port 8931');
  });
  */

  // For simpler non-web transport:
  // server.connect(yourCustomTransport);

  console.log("Playwright MCP server started programmatically.");

  // Keep the server running and handle connections.
  // Implement server shutdown logic.
}

runMyMCPServer().catch(console.error);
```

### Response
#### Success Response (N/A - Script execution)
N/A

#### Response Example
N/A
```

--------------------------------

### Install Project Dependencies

Source: https://www.claudemcp.com/servers/en/image-gen-server

This section provides commands for cloning the project repository and installing Python dependencies using pip and uv. It also outlines the necessary environment preparations, including Python, npm, and Node.js versions.

```bash
git clone https://github.com/fengin/image-gen-server.git
cd image-gen-server
pip install -r requirements.txt
pip install uv
```

--------------------------------

### Install Development Dependencies with uv

Source: https://www.claudemcp.com/servers/en/seo-mcp

Installs SEO MCP in editable mode for development using the `uv` package manager. This is an alternative to using pip.

```bash
uv sync
```

--------------------------------

### Install iterm-mcp via Smithery (Bash)

Source: https://www.claudemcp.com/servers/en/iterm-mcp

This command installs the iterm-mcp server for Claude Desktop automatically using the Smithery CLI. It simplifies the setup process by handling dependencies and configuration. Ensure you have npx installed and available in your PATH.

```bash
npx -y @smithery/cli install iterm-mcp --client claude
```

--------------------------------

### Windows winget을 사용한 필수 도구 설치

Source: https://www.claudemcp.com/docs/ko/quickstart

Windows 시스템에 Claude MCP 프로토콜을 사용하기 위한 필수 도구인 uv, Git, SQLite를 winget을 사용하여 설치합니다. 이러한 도구는 Claude 데스크톱이 로컬 데이터베이스와 상호 작용하는 데 필요합니다.

```bash
# Using winget
winget install --id=astral-sh.uv -e
winget install git.git sqlite.sqlite
```

--------------------------------

### Clone and Install SEO MCP Locally

Source: https://www.claudemcp.com/servers/tw/seo-mcp

These commands clone the SEO MCP repository from GitHub and install the project locally using pip. This is useful for development or manual installation. It requires Git and Python 3.10+.

```bash
git clone https://github.com/cnych/seo-mcp.git
cd seo-mcp
pip install -e .
```

--------------------------------

### Run Bright Data MCP Server

Source: https://www.claudemcp.com/servers/en/brightdata

These examples demonstrate how to initiate the Bright Data MCP server from the command line. The first example uses Node.js package runner (npx) with the necessary API token environment variable. The second shows how to optionally specify a custom browser zone.

```shell
# Basic execution with API Token
API_TOKEN=<your-token> npx @brightdata/mcp
```

```shell
# Execution with custom Browser Zone
API_TOKEN=<your-token> BROWSER_ZONE=<zone-name> npx @brightdata/mcp
```

--------------------------------

### SQLite 데이터베이스 생성 및 데이터 삽입

Source: https://www.claudemcp.com/docs/ko/quickstart

Claude MCP 프로토콜 예제를 위해 'products' 테이블이 있는 간단한 SQLite 데이터베이스를 생성하고 샘플 데이터를 삽입합니다. 이 데이터베이스는 Claude 데스크톱이 쿼리할 로컬 데이터 소스로 사용됩니다.

```bash
# Create a new SQLite database
sqlite3 ~/test.db <<EOF
CREATE TABLE products (
  id INTEGER PRIMARY KEY,
  name TEXT,
  price REAL
);

INSERT INTO products (name, price) VALUES
  ('Widget', 19.99),
  ('Gadget', 29.99),
  ('Gizmo', 39.99),
  ('Smart Watch', 199.99),
  ('Wireless Earbuds', 89.99),
  ('Portable Charger', 24.99),
  ('Bluetooth Speaker', 79.99),
  ('Phone Stand', 15.99),
  ('Laptop Sleeve', 34.99),
  ('Mini Drone', 299.99),
  ('LED Desk Lamp', 45.99),
  ('Keyboard', 129.99),
  ('Mouse Pad', 12.99),
  ('USB Hub', 49.99),
  ('Webcam', 69.99),
  ('Screen Protector', 9.99),
  ('Travel Adapter', 27.99),
  ('Gaming Headset', 159.99),
  ('Fitness Tracker', 119.99),
  ('Portable SSD', 179.99);
EOF
```

--------------------------------

### Install Firecrawl MCP via Smithery CLI

Source: https://www.claudemcp.com/servers/en/firecrawl-mcp

This command uses the Smithery CLI to automatically install the Firecrawl MCP server for Claude Desktop, simplifying the setup process for that specific client.

```bash
npx -y @smithery/cli install @mendableai/mcp-server-firecrawl --client claude
```

--------------------------------

### Install mcp-server-webcrawl using pip

Source: https://www.claudemcp.com/servers/en/mcp-server-webcrawl

This command installs the mcp-server-webcrawl package using pip. Ensure you have Python 3.10 or higher. This is the primary method for setting up the server.

```bash
pip install mcp-server-webcrawl
```

--------------------------------

### Manual Installation Configuration (uvx)

Source: https://www.claudemcp.com/servers/ko/dify-mcp-server

This JSON configuration is for installing and running the Dify MCP server using 'uvx'. This method is recommended as it allows usage without cloning the code locally. It specifies the command, the source Git repository, and the environment variable for the configuration path.

```json
{
  "mcpServers": {
    "dify-mcp-server": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/YanxingLiu/dify-mcp-server",
        "dify_mcp_server"
      ],
      "env": {
        "CONFIG_PATH": "/Users/lyx/Downloads/config.yaml"
      }
    }
  }
}
```

--------------------------------

### Initialize npm Project and Install Dependencies (Bash)

Source: https://www.claudemcp.com/docs/en/write-ts-client

Commands to create a new project directory, initialize an npm project, and install necessary Node.js dependencies and TypeScript development tools.

```bash
mkdir mcpclient-demo
cd mcpclient-demo
npm init -y
npm install @anthropic-ai/sdk @modelcontextprotocol/sdk dotenv
npm install -D @types/node typescript
```

--------------------------------

### Clone Repository and Install Dependencies

Source: https://www.claudemcp.com/servers/en/perplexity-ask-mcp-server

Clones the modelcontextprotocol repository and installs npm dependencies for the perplexity-ask directory.

```bash
git clone git@github.com:ppl-ai/modelcontextprotocol.git
cd modelcontextprotocol/perplexity-ask && npm install
```

--------------------------------

### macOS Homebrew를 사용한 필수 도구 설치

Source: https://www.claudemcp.com/docs/ko/quickstart

macOS 시스템에 Claude MCP 프로토콜을 사용하기 위한 필수 도구인 uv, Git, SQLite를 Homebrew를 사용하여 설치합니다. 이러한 도구는 Claude 데스크톱이 로컬 데이터베이스와 상호 작용하는 데 필요합니다.

```bash
# Using Homebrew 사용
brew install uv git sqlite3
```

--------------------------------

### Install MCP K8S Go with Smithery CLI

Source: https://www.claudemcp.com/servers/en/mcp-k8s-go

This command installs the MCP K8S Go server for Claude Desktop automatically using the Smithery CLI tool.

```bash
npx -y @smithery/cli install @strowk/mcp-k8s --client claude
```

--------------------------------

### Start Playwright MCP Server with SSE

Source: https://www.claudemcp.com/servers/en/playwright

This bash command starts the Playwright MCP server using SSE transport on port 8931. It allows the server to run in the background on systems without a graphical interface. The client can then connect to this server using the specified URL.

```bash
npx @playwright/mcp --port 8931
```

--------------------------------

### Install and Configure Context7 MCP Server (JSON)

Source: https://context7

Shows how to install and configure the Context7 MCP server using npx. This server fetches version-specific documentation for libraries into LLM context. Configuration can be done in ~/.cursor/mcp.json or Claude Desktop configuration.

```json
// Cursor configuration: ~/.cursor/mcp.json
{
  "mcpServers": {
    "context7": {
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp@latest"]
    }
  }
}

// Claude Desktop configuration
{
  "mcpServers": {
    "Context7": {
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp@latest"]
    }
  }
}

// Claude Code
// Run: claude mcp add context7 -- npx -y @upstash/context7-mcp@latest
```

--------------------------------

### Start Weather Server Command

Source: https://www.claudemcp.com/blogs/en/langchain-mcp-adapters

This command initiates the Python weather server script. It displays the server startup logs, indicating that the Uvicorn server has started and is listening for connections on http://0.0.0.0:8000.

```bash
$python weather_server.py
INFO:     Started server process [64550]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)
```

--------------------------------

### Python Client Demo: Load MCP Tools and Create LangChain Agent

Source: https://www.claudemcp.com/blogs/ko/langchain-mcp-adapters

This Python script demonstrates how to set up an MCP client session, load tools from an MCP server using `load_mcp_tools`, and then create a LangChain React agent with these tools. It showcases initializing the client, loading tools, and invoking the agent with a query, printing all messages including tool calls and responses. Dependencies include `mcp`, `langchain_mcp_adapters`, and `langgraph`.

```python
from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import stdio_client
import asyncio

from langchain_mcp_adapters.tools import load_mcp_tools
from langgraph.prebuilt import create_react_agent

from langchain_openai import ChatOpenAI
model = ChatOpenAI(model="gpt-4o")

server_params = StdioServerParameters(
    command="python",
    # math_server.py 의 전체 절대 경로를 확인하세요.
    args=["/your/path/to/math_server.py"],
)

async def main():
    async with stdio_client(server_params) as (read, write):
        async with ClientSession(read, write) as session:
            # 초기화 연결
            await session.initialize()

            # 도구 가져오기
            tools = await load_mcp_tools(session)
            print(f"tools: {tools}")
            # 에이전트 생성 및 실행
            agent = create_react_agent(model, tools)
            agent_response = await agent.ainvoke({"messages": "what's (3 + 5) x 12?"})

            # 모든 메시지 출력
            print("All messages:")
            for message in agent_response["messages"]:
                print(f"Message type: {type(message).__name__}")
                print(f"Message content: {message.content}")
                if hasattr(message, 'tool_calls') and message.tool_calls:
                    print(f"Tool calls: {message.tool_calls}")
                if hasattr(message, 'name') and message.name:
                    print(f"Tool name: {message.name}")
                if hasattr(message, 'tool_call_id') and message.tool_call_id:
                    print(f"Tool call id: {message.tool_call_id}")
                print("-" * 50)

if __name__ == "__main__":
    asyncio.run(main())

```

--------------------------------

### Develop Cloudflare MCP Server Locally

Source: https://www.claudemcp.com/servers/en/claudflare

Commands to install dependencies and build the Cloudflare MCP Server in watch mode for local development. A second terminal is needed to start the server after building.

```bash
pnpm install
pnpm build:watch
```

--------------------------------

### Keyword Generator API Call (Python Example)

Source: https://www.claudemcp.com/servers/ko/seo-mcp

This Python code snippet demonstrates how to use the `keyword_generator` MCP tool to get keyword and question ideas for a specific keyword. It shows how to specify the keyword, country, and search engine.

```python
from seo_mcp import MCP
mcp = MCP()

keyword_to_search = "best seo tools"

keyword_ideas = mcp.keyword_generator(keyword=keyword_to_search, country="us", search_engine="Google")
print(keyword_ideas)
```

--------------------------------

### Configure Filesystem MCP Server in Claude Desktop

Source: https://www.claudemcp.com/docs/en/quickstart

This JSON configuration adds the filesystem MCP server to Claude Desktop. It specifies the command to run the server and its arguments, including allowed directory paths. Ensure paths are updated to your local environment.

```json
{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "/Users/username/Desktop",
        "/path/to/other/allowed/dir"
      ]
    }
  }
}
```

--------------------------------

### Install MCP K8S Go Globally via npm

Source: https://www.claudemcp.com/servers/en/mcp-k8s-go

This command installs the pre-built binaries for MCP K8S Go globally using npm, allowing it to be run from the command line.

```bash
npm install -g @strowk/mcp-k8s
```

--------------------------------

### Run iterm-mcp with Auto-Rebuild for Development (Bash)

Source: https://www.claudemcp.com/servers/en/iterm-mcp

This command starts the iterm-mcp server in watch mode, which automatically rebuilds the server whenever code changes are detected. This is useful for rapid development and testing. Ensure Node.js and Yarn are installed.

```bash
yarn run watch
```

--------------------------------

### Get Backlinks List API Call (Python Example)

Source: https://www.claudemcp.com/servers/ko/seo-mcp

This Python code snippet demonstrates how to call the `get_backlinks_list` MCP tool to retrieve backlink data for a given domain. It shows the function call and expects a list of backlink objects as a response.

```python
from seo_mcp import MCP
mcp = MCP()

domain_to_check = "example.com"
backlinks = mcp.get_backlinks_list(domain=domain_to_check)

print(backlinks)
```

--------------------------------

### Setting up MCP Server

Source: https://www.claudemcp.com/servers/zh/playwright

Instructions for starting the MCP server on a system without a graphical interface.

```APIDOC
## Starting the MCP Server

To run the Playwright MCP server on a Linux system without a display or for IDE work processes, use the following command:

### Command
```bash
npx @playwright/mcp --port 8931
```

### Configuration
Once the server is running, configure the MCP client with the following JSON:

### Request Example
```json
{
  "mcpServers": {
    "playwright": {
      "url": "http://localhost:8931/sse"
    }
  }
}
```
```

--------------------------------

### Backlinks MCP Tool: Get Backlinks List

Source: https://www.claudemcp.com/servers/en/backlinks-mcp

An example of how to call the `get_backlinks_list` tool provided by the Backlinks MCP. This tool takes a domain name as input and returns a list of backlink objects.

```python
from backlinks_mcp import get_backlinks_list

domain = "example.com"
backlinks = get_backlinks_list(domain)
print(backlinks)

```

--------------------------------

### Prompting with Context7 MCP in Cursor

Source: https://www.claudemcp.com/servers/zh/context7

This example shows how to use the Context7 MCP server within the Cursor IDE. By appending 'use context7' to a prompt, Cursor will leverage Context7 to fetch and inject the most current documentation and code examples relevant to the request.

```txt
创建一个使用 app router 的基础 Next.js 项目。use context7
```

```txt
编写根据 PostgreSQL 凭证删除 city 列为空值的脚本。use context7
```

--------------------------------

### CLI Installation for Magic AI Agent

Source: https://www.claudemcp.com/servers/en/magic-mcp

Installs and configures the Magic Component Platform (MCP) for a specified client using the CLI. Requires an API key for authentication. Supported clients include cursor, windsurf, cline, and claude.

```bash
npx @21st-dev/cli@latest install <client> --api-key <key>
```

--------------------------------

### SEO MCP API - Get Backlinks List (Python Example)

Source: https://www.claudemcp.com/servers/tw/seo-mcp

This Python function demonstrates how to call the `get_backlinks_list` MCP tool to retrieve backlink data for a given domain. It assumes the SEO MCP service is running and accessible.

```python
def get_backlinks_list(domain: str):
    """獲取指定網域的反向連結列表。"""
    pass  # 實際實現會調用 MCP 服務
```

--------------------------------

### Create project directory and init npm

Source: https://www.claudemcp.com/docs/zh/write-ts-client

Creates a directory for the MCP client demo and initializes an npm project. This sets up the basic project structure for the client application.

```bash
mkdir mcpclient-demo
cd mcpclient-demo
npm init -y
```

--------------------------------

### SEO MCP API - Keyword Generator (Python Example)

Source: https://www.claudemcp.com/servers/tw/seo-mcp

This Python function shows how to use the `keyword_generator` MCP tool to get keyword and SEO suggestions for a given keyword, country, and search engine. It assumes the SEO MCP service is running.

```python
def keyword_generator(keyword: str, country: str = "us", search_engine: str = "Google"):
    """獲取指定關鍵字的創意和 SEO 建議。"""
    pass  # 實際實現會調用 MCP 服務
```

--------------------------------

### Dockerfile for Context7 MCP Server

Source: https://www.claudemcp.com/servers/en/context7

This Dockerfile provides instructions for building a Docker image to run the Context7 MCP server. It sets up a Node.js environment, installs the MCP server globally, and defines the command to run it.

```Dockerfile
FROM node:18-alpine

WORKDIR /app

RUN npm install -g @upstash/context7-mcp@latest

CMD ["context7-mcp"]
```

--------------------------------

### Install Dify MCP Server using Smithery CLI

Source: https://www.claudemcp.com/servers/en/dify-mcp-server

This command installs the Dify MCP Server using the Smithery CLI. It specifies the package name 'dify-mcp-server' and the client as 'claude'. Other supported clients include 'cline, windsurf, roo-cline, witsy, enconvo, cursor'.

```bash
npx -y @smithery/cli install dify-mcp-server --client claude
```

--------------------------------

### Example Log Messages (Text)

Source: https://www.claudemcp.com/servers/tw/firecrawl-mcp

Sample log messages generated by the Firecrawl MCP Server, illustrating various events such as initialization, scraping start, batch operation queuing, credit usage warnings, and rate limit errors. These logs are essential for monitoring server status and debugging.

```text
[INFO] Firecrawl MCP Server 成功初始化
[INFO] 開始爬取 URL: https://example.com
[INFO] 批次操作已加入佇列，ID: batch_1
[WARNING] 信用額度使用已達到警告閾值
[ERROR] 超過速率限制，將在 2 秒後重試...
```

--------------------------------

### Install uvx via Command Line

Source: https://www.claudemcp.com/servers/en/PixVerse-MCP

This snippet provides commands to copy the uvx executable to a system's PATH for resolving spawn uvx ENOENT errors. It includes separate instructions for Mac/Linux and Windows.

```bash
sudo cp ./uvx /usr/local/bin
```

```batch
where uvx

```

--------------------------------

### Python LangGraph Agent Setup with MCP Adapters

Source: https://www.claudemcp.com/blogs/zh/langchain-mcp-adapters

This Python code sets up an asynchronous context manager to create a LangGraph agent. It initializes an MCP client to connect to multiple servers (math and weather) and retrieves their tools to be used by the LangChain agent. The `create_react_agent` function from `langgraph.prebuilt` is used to build the agent.

```python
from contextlib import asynccontextmanager
from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.prebuilt import create_react_agent
from langchain_anthropic import ChatAnthropic

model = ChatAnthropic(model="claude-3-5-sonnet-latest")

@asynccontextmanager
async def make_graph():
    async with MultiServerMCPClient(
        {
            "math": {
                "command": "python",
                # 确保更新到 math_server.py 的完整绝对路径
                "args": ["/path/to/math_server.py"],
                "transport": "stdio",
            },
            "weather": {
                # 确保你从 weather_server.py 开始
                "url": "http://localhost:8000/sse",
                "transport": "sse",
            }
        }
    ) as client:
        agent = create_react_agent(model, client.get_tools())
        yield agent
```

--------------------------------

### Install FastMCP with uv

Source: https://www.claudemcp.com/servers/ko/backlinks-mcp

This code snippet demonstrates how to install the FastMCP library using the 'uv' package manager. 'uv' is a fast Python package installer and resolver. Ensure you have Python 3.8+ and 'uv' installed before running this command.

```bash
uv pip install fastmcp
```

--------------------------------

### Check Node.js Installation

Source: https://www.claudemcp.com/servers/ko/supabase-mcp

Verify if Node.js is installed on your system. This is a prerequisite for setting up the Supabase MCP Server. If not installed, download it from nodejs.org.

```shell
node -v
```

--------------------------------

### Install Development Dependencies

Source: https://www.claudemcp.com/servers/zh/seo-mcp

Installs development dependencies for the SEO MCP project after cloning the repository. It uses 'uv sync' for efficient dependency management, or alternatively 'pip install -e .'. Ensure Git and Python 3.10+ are installed.

```bash
git clone https://github.com/cnych/seo-mcp.git
cd seo-mcp
uv sync  # or use pip install -e .
```

--------------------------------

### 創建 SQLite 資料庫並插入數據

Source: https://www.claudemcp.com/docs/tw/quickstart

此代碼片段創建了一個名為 `test.db` 的 SQLite 資料庫，並在其中創建了一個 `products` 表，然後插入了 20 條產品數據。這是設置 Claude MCP 與本地資料庫交互的準備步驟。

```bash
sqlite3 ~/test.db <<EOF
CREATE TABLE products (
  id INTEGER PRIMARY KEY,
  name TEXT,
  price REAL
);

INSERT INTO products (name, price) VALUES
  ('Widget', 19.99),
  ('Gadget', 29.99),
  ('Gizmo', 39.99),
  ('Smart Watch', 199.99),
  ('Wireless Earbuds', 89.99),
  ('Portable Charger', 24.99),
  ('Bluetooth Speaker', 79.99),
  ('Phone Stand', 15.99),
  ('Laptop Sleeve', 34.99),
  ('Mini Drone', 299.99),
  ('LED Desk Lamp', 45.99),
  ('Keyboard', 129.99),
  ('Mouse Pad', 12.99),
  ('USB Hub', 49.99),
  ('Webcam', 69.99),
  ('Screen Protector', 9.99),
  ('Travel Adapter', 27.99),
  ('Gaming Headset', 159.99),
  ('Fitness Tracker', 119.99),
  ('Portable SSD', 179.99);
EOF
```