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)
Create videos with prompts, and use Rendley’s video editor to adjust anything you need.
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。
Authorization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 …”。
这个地址需要在请求的 header 中放入密钥,而添加连接器的菜单无法设置。请改用 Claude Code、Cursor 或 VS Code。
在终端中运行这条命令
claude mcp add --transport http rendley https://mcp.rendley.com/mcp --header 'Authorization: Bearer {rendley_api_key}'运行前请先填写空着的值。
粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"rendley": {
"url": "https://mcp.rendley.com/mcp",
"headers": {
"Authorization": "Bearer {rendley_api_key}"
}
}
}
}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": {
"rendley": {
"type": "http",
"url": "https://mcp.rendley.com/mcp",
"headers": {
"Authorization": "Bearer {rendley_api_key}"
}
}
}
}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。
这个地址需要在请求的 header 中放入密钥,而添加连接器的菜单无法设置。请改用 Claude Code、Cursor 或 VS Code。
你的选择通过 Cookie 保存在这台设备上,无需账户。
Authorization HTTP header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。
我们还没有读取这个服务器的工具列表;我们会轮流检查各个端点。
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8日
注册名称已通过 GitHub 账户 rendleyhq 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
尚未检查;我们会轮流检查各个端点
尚未检查;我们会轮流检查各个端点
尚未检查;我们会在检查端点时读取工具列表
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。
它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。
GitHub 仓库最后更新于–,最新版本(1.0.1)于5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。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。
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:↓ 每周下载量i★ 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。
The MCP server for Rendley. It gives an AI assistant a full video editor: connect it once, then create and edit video by describing what you want, in the same chat you already work in. Under the hood it drives the hosted Rendley editor in a browser and calls the hosted Rendley API on your behalf:
https://app.rendley.comhttps://api.rendley.com/v1The same agent is also reachable over plain HTTP (POST /v1/agent) for backends that don't speak MCP. See REST example.
There are two ways to use it:
https://mcp.rendley.com/mcp. Nothing to run. See Connect a client.📖 Full documentation, with step-by-step setup guides and screenshots: [docs.rendley.com/mcp](https://docs.rendley.com/mcp)
Any client that speaks remote MCP can connect. Point it at the endpoint and authenticate by signing in or with an API key:
https://mcp.rendley.com/mcp
Authorization: Bearer YOUR_RENDLEY_API_KEY header. Create one at Settings → API Keys; see Get an API key.Most clients accept a JSON config like this:
{
"mcpServers": { "rendley": { "url": "https://mcp.rendley.com/mcp", "headers": { "Authorization": "Bearer YOUR_RENDLEY_API_KEY" } } } }
For sign-in, drop the headers block and let the client run the browser flow.
The docs have illustrated, per-client walkthroughs:
mcp-remote bridge for local-only clients.For Claude Code specifically:
claude mcp add --transport http rendley https://mcp.rendley.com/mcp \
--header "Authorization: Bearer YOUR_RENDLEY_API_KEY"
Ask the client what it can do, then try a simple request:
What can Rendley do here?
List my projects.
If it lists the Rendley tools and your projects, you're connected. From there, describe the video you want and the assistant creates a project, builds the edit, and returns a link.
This repo is the MCP server only; it does not run the full Rendley stack locally. It talks to the hosted Rendley API and editor. When you self-host, point clients at your own https:// /mcp instead of https://mcp.rendley.com/mcp.
>= 1.1 (or Docker)git clone https://github.com/rendleyhq/rendley-mcp.git
cd rendley-mcp
bun install bun dev # http://localhost:8787
API_BASE_URL and APP_BASE_URL default to the hosted Rendley API and editor, so the server boots out of the box with no .env at all; copy .env.example only to point at a different stack. The server holds no credentials of its own: every request authenticates with the caller's own API key or OAuth token.
By default the server drives the editor with a local Chrome (BROWSER_MODE=local), so no extra infrastructure is needed. bun install downloads a bundled Chromium; with USE_CHROME_CHANNEL=true it prefers your installed Google Chrome and falls back to the bundled browser. Set HEADLESS=false to watch the editor in a visible window, and CPU_ONLY=true on machines without a usable GPU.
Verify it's up:
curl http://localhost:8787/health
docker compose up -d --build
curl http://localhost:8787/health
Use the same client steps as the hosted server, but swap the endpoint for your host (e.g. http://localhost:8787/mcp) and authenticate with your API key as a bearer token:
{
"mcpServers": { "rendley": { "url": "http://localhost:8787/mcp", "headers": { "Authorization": "Bearer YOUR_RENDLEY_API_KEY" } } } }
OAuth. OAuth sign-in is always on. Clients that support it (Claude connectors, ChatGPT, codex mcp login rendley) can sign in via the hosted login flow instead of pasting a key; the server advertises the authorization server through RFC 9728 protected-resource metadata. Set MCP_PUBLIC_URL to this server's public URL so discovery points at the right host.
For hosted deployments, set BROWSER_MODE=remote and point the server at a deployed remote browser worker. The server mints a short-lived editor session token and posts it to the worker, so the caller's credential never leaves this process:
BROWSER_MODE=remote
BROWSER_WORKER_URL=https:// BROWSER_WORKER_TOKEN=
All configuration is via environment variables; start from .env.example, which documents every option. The essentials:
| Variable | Default | Purpose |
|---|---|---|
API_BASE_URL | https://api.rendley.com/v1 | Rendley API root. Override for staging or a self-hosted stack. |
APP_BASE_URL | https://app.rendley.com | Rendley editor root. Override for staging or a self-hosted stack. |
MCP_PUBLIC_URL | none | This server's public URL, advertised for OAuth discovery. |
BROWSER_MODE | local | local = in-process Playwright Chrome; remote = remote browser worker. |
HEADLESS | true | Local mode only; false shows the browser window. |
USE_CHROME_CHANNEL | true | Local mode only; prefer installed Chrome over bundled Chromium. |
CPU_ONLY | false | Force software WebGL (SwiftShader) on hosts without a GPU. |
QUEUE_CONCURRENCY | 120 | Max concurrent browser-backed jobs per container. |
CORS_ORIGINS | none | CSV allowlist of browser origins. Empty disables CORS. |
| Method | Path | |
|---|---|---|
POST | /mcp | MCP Streamable HTTP transport |
POST | /v1/agent | Start an async edit job |
GET | /v1/jobs/:id | Poll a job |
GET | /health | Liveness |
You don't need an MCP client to drive the agent. The same agent is available over plain HTTP. Start a job, get an id back, and poll until the video is ready. Full reference: docs.rendley.com/mcp/api-access.
# Start a job (swap in http://localhost:8787 for a self-hosted instance)
curl -X POST https://mcp.rendley.com/v1/agent \ -H "Authorization: Bearer $RENDLEY_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "prompt": "Create a 9:16 slideshow from these files", "files": [ { "url": "https://cdn.example.com/photo1.jpg" }, { "url": "https://cdn.example.com/photo2.jpg" } ] }'
curl -H "Authorization: Bearer $RENDLEY_API_KEY" \ https://mcp.rendley.com/v1/jobs/
/health is healthy whenever the server can serve; browser launch failures surface on browser-backed requests.bun dev # watch mode, pretty logs
bun start # production start bun run typecheck
Apache License 2.0. © 2026 Rendley.