Cursor Talk to Figma MCP
Enables Cursor AI to read and modify Figma designs programmatically via the Model Context Protocol.
README
Cursor Talk to Figma MCP
Note: This repository was created solely to make the MCP integration work via the
npxcommand instead ofbunxfor easier compatibility and setup.
A Model Context Protocol (MCP) integration that allows Cursor AI to communicate with Figma for reading and modifying designs programmatically.
Prerequisite:
You need Node.js (which includes npx) installed on your system. Download and install it from the official Node.js website. After installation, make sure the installation path for Node.js and npm is added to your system's environment PATH variable so you can run node, npm, and npx from any terminal window.
Default installation paths:
- Windows:
C:\Program Files\nodejs\- macOS/Linux:
/usr/local/bin/(or wherever your package manager installs global binaries)
You can check your versions with:
node -v
npx -v
Screenshots:
-

Figma plugin with 'Use localhost' enabled
-

Running MCP via npx in terminal
-

MCP server running via npx
-

MCP server running via cmd and npx on Windows
Installation via NPX
Follow these steps to install and set up the Cursor Talk to Figma MCP:
1. Start the WebSocket Server
The WebSocket server facilitates communication between Cursor and Figma.
- Download the
node_socket_server.mjsfile to your computer - Open a terminal/command prompt and navigate to the directory containing the file
- Start the server with Node.js:
node node_socket_server.mjs
Alternatively, create a batch file (Windows) or shell script (Mac/Linux) for easy startup.
2. Configure Cursor MCP Server
- In Cursor, navigate to Settings > MCPServers
- Add a new MCP server configuration:
{
"mcpServers": {
"TalkToFigma": {
"command": "npx",
"args": [
"cursor-talk-to-figma-mcp@latest"
]
}
}
}
Tip: You can also run this MCP using
cmdfor compatibility on Windows systems. Here is an examplemcpServersconfiguration for Cursor:{ "mcpServers": { "TalkToFigma-mcp": { "command": "cmd", "args": [ "bash", "-c", "cmd /c npx -y cursor-talk-to-figma-mcp@latest" ] } } }This allows you to invoke the MCP server through
cmdusingnpx, which is especially useful if you encounter compatibility issues with other shells or environments on Windows.
3. Install the Figma Plugin
Option 1: Install from Figma Community
- Visit the Cursor Talk to Figma MCP Plugin on Figma Community
- Click "Install" to add it to your Figma plugins
Option 2: Install Locally
- In Figma, go to Plugins > Development > New Plugin
- Choose "Link existing plugin"
- Select the
manifest.jsonfile from thesrc/cursor_mcp_plugindirectory - The plugin will now appear in your Figma development plugins
4. Connect Cursor to Figma
- Open Figma and run the Cursor Talk To Figma MCP Plugin
- In the plugin panel, ensure "Use localhost" is enabled on 3055 port
- Note the channel name shown in the green connected status message (e.g., "959ykpvn")
- In Cursor, join the channel by typing:
mcp_TalkToFigma-mcp_join_channel with channel: "[CHANNEL_NAME]"
Replace [CHANNEL_NAME] with the channel name from the Figma plugin.
Usage
After connecting, you can use any of the available MCP commands in Cursor to interact with your Figma design. Here are some examples:
mcp_TalkToFigma-mcp_get_document_info
Gets information about the current Figma document
mcp_TalkToFigma-mcp_get_selection
Gets information about the currently selected elements
mcp_TalkToFigma-mcp_create_rectangle with x: 100, y: 100, width: 200, height: 100
Creates a new rectangle at the specified position and size
mcp_TalkToFigma-mcp_create_text with x: 100, y: 100, text: "Hello World"
Creates a new text element with the specified text
Available Commands
The MCP integration provides a wide range of commands for:
- Reading document and selection information
- Creating and modifying design elements
- Working with text, colors, and styles
- Managing auto-layout properties
- Handling component instances and overrides
- Creating and managing annotations
- Exporting designs as images
For a complete list of available commands and their parameters, refer to the full documentation.
Troubleshooting
- If you see "Must join a channel" errors, make sure you've joined the correct channel from the Figma plugin
- If the WebSocket server disconnects, restart it and reconnect the plugin
- If Cursor can't find the MCP server, verify your MCPServers configuration
- Ensure that port 3055 is available for the WebSocket server
License
Credits
All credits go to sonnylazuardi/cursor-talk-to-figma-mcp, as this project was cloned from their repository: https://github.com/sonnylazuardi/cursor-talk-to-figma-mcp.git
This fork was made solely to enable MCP usage via npx instead of bunx.
This allows you to invoke the MCP server through cmd using npx, which is especially useful if you encounter compatibility issues with other shells or environments on Windows.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。