一个用于将Pinterest API与AI代理和大型语言模型集成的模型上下文协议（MCP）服务器。

概述

该MCP服务器允许AI代理以标准化方式与Pinterest API交互，支持以下功能：

• 检索用户信息
• 管理Pinterest板块（列表、创建、浏览）
• 管理图钉（列表、创建、查看）

先修条件

• Node.js（v14或更新版本）
• Pinterest 开发者账户
• Pinterest API 应用凭证

布置

1. 克隆此仓库：

2. 安装依赖：

3. 从模板创建 .env 文件：

4. 用你的Pinterest API凭证编辑.env文件：

5. 在Pinterest开发者平台注册你的申请

• 添加 http://localhost:3052/ 作为授权重定向 URI

用途

1. 启动MCP服务器：

2. 如果不存在认证令牌，服务器会自动打开浏览器窗口进行认证。

3. 认证成功后，MCP服务器将准备好接受来自AI代理的连接。

与人工智能代理连接

MCP 服务器在 Socket.IO 上实现 MCP 协议，并公开以下工具：

• pinterest.user.get_info - 获取用户信息

Pinterest MCP Server

A Model Context Protocol (MCP) server for integrating the Pinterest API with AI agents and LLMs.

Overview

This MCP server allows AI agents to interact with the Pinterest API in a standardized way, enabling functionalities like:

• Retrieving user information
• Managing Pinterest boards (listing, creating, viewing)
• Managing pins (listing, creating, viewing)

Prerequisites

• Node.js (v14 or newer)
• Pinterest Developer Account
• Pinterest API App credentials

Setup

1. Clone this repository:

git clone https://github.com/yourusername/pinterest-mcp-server.git
cd pinterest-mcp-server

2. Install dependencies:

npm install

3. Create a .env file from the template:

cp .env.template .env

4. Edit the .env file with your Pinterest API credentials:

PINTEREST_APP_ID=your_pinterest_app_id
PINTEREST_APP_SECRET=your_pinterest_app_secret

5. Register your application on the Pinterest Developer Platform
   • Add http://localhost:3052/ as an authorized redirect URI

Usage

1. Start the MCP server:

npm start

2. If no authentication token exists, the server will automatically open a browser window for authentication.

3. After successful authentication, the MCP server will be ready to accept connections from AI agents.

Connecting with AI Agents

The MCP server implements the MCP protocol over Socket.IO and exposes the following tools:

• pinterest.user.get_info - Get user information
• pinterest.boards.list - List user’s boards
• pinterest.boards.create - Create a new board
• pinterest.boards.get - Get board details
• pinterest.pins.list - List pins on a board
• pinterest.pins.create - Create a new pin
• pinterest.pins.get - Get pin details

You can connect your AI agents to this MCP server using compatible MCP client libraries.

Example (Claude.ai Desktop App)

1. Make sure your MCP server is running
2. Open Claude Desktop App
3. Click on “Connect local tools” (bottom right)
4. Enter http://localhost:3052 as the server URL
5. You can now ask Claude to perform Pinterest actions like:
   • “List my Pinterest boards”
   • “Create a new Pinterest board called ‘Vacation Ideas’”
   • “Pin this image to my Travel board”

License

MIT