Figma Context MCP
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
作者提供的说明(英文原文)
One audio queue across Spotify, the browser, local files and whatever is already playing.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"upnext": {
"command": "npx",
"args": [
"-y",
"upnext-mcp"
]
}
}
}在终端中运行这条命令
claude mcp add upnext -- npx -y upnext-mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"upnext": {
"command": "npx",
"args": [
"-y",
"upnext-mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"upnext": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"upnext-mcp"
]
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月7日
注册名称已通过 GitHub 账户 tothienbao6a0 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(0.3.1)于2个月前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
同一分类中有:Figma Context MCP, Claude Real Video, Figma MCP Server
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
Let any LLM watch a video locally — and search everything it has ever watched.
The Figma MCP server brings Figma design context directly into your AI workflow.
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
An MCP server for your machine's audio. One queue across Spotify, the browser, local files, and whatever is already playing.
{
"mcpServers": { "upnext": { "command": "npx", "args": ["-y", "upnext-mcp", "--library", "/Users/me/Music"] } } }
Drop that in Claude Desktop, Cursor, Windsurf, or anything else that speaks MCP.
you: what am I listening to?
→ ❚❚ Acquired — Jensen Huang (Google Chrome)
you: let that finish, then play Bad Habit → adopted "Acquired — Jensen Huang" as q_0001 up next: q_0001 Acquired — Jensen Huang q_0002 Bad Habit — Steve Lacy
The first line reads a browser tab — no extension. The second puts it in the queue as a real entry, so the agent adds to what someone is doing rather than cutting across it.
| tool | |
|---|---|
media_now | what's playing, in the queue *and* elsewhere on the machine |
media_queue | the queue, with the id of each entry |
media_enqueue | add a link, file path, Spotify URI, or a plain description |
media_play · media_pause · media_next | transport |
media_seek | jump to a position — says so plainly if the source can't |
media_move · media_remove | reorder and drop, by id |
media_search | search the sources that can search |
media_adopt_current | put what's already playing into the queue |
media_setup | what's wired here, and what it can do |
Every mutation is addressed by id, and every response includes the queue with ids in it. That's deliberate: a model working from positions races anything else touching the queue, and by the time its call lands, position 2 is a different song.
Two places where this behaves differently from most tool servers, both for the same reason — a model that gets a silent no-op will confidently tell the user it worked.
`media_seek` on a source that can't seek returns a sentence, not an error:
the current source (spotify-desktop) cannot seek. Nothing changed.
`media_search` with nothing wired to search explains itself:
nothing found for "bad habit".
playing through: spotify-desktop, nowplaying, local
titles will NOT resolve: nothing wired here can search…
A link says exactly what to play. A *description* has to be looked up, and not every backend can. The Spotify desktop app plays a URI you hand it but cannot search a catalogue — so on a plain Mac setup, media_enqueue with "bad habit" has nothing to resolve against.
Give it one of these:
// index a music folder
"args": ["-y", "upnext-mcp", "--library", "/Users/me/Music"]
// or a Spotify Web token, via the environment "env": { "SPOTIFY_TOKEN": "..." }
media_setup will tell you which you have.
If your product already has a queue, expose *that* one rather than starting a second that competes with it:
import { createServer } from 'upnext-mcp';
import { serveStdio } from '@modelcontextprotocol/server/stdio';
serveStdio(() => createServer(myExistingRuntime));
This package is a transport, and nothing more — ordering, capability negotiation, resolution and reconciliation all happen in upnext-core. It lives apart because a transport is an opinion, and nobody importing the runtime should inherit this one.
Full docs: https://github.com/tothienbao6a0/upnext · Apache-2.0