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)
License-first federated image search with attribution, dimensions, and download tooling.
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。
UNSPLASH_ACCESS_KEY, PEXELS_API_KEY, PIXABAY_API_KEY, BRAVE_API_KEY, SERPAPI_KEY标为机密的值是你账户的钥匙。千万不要分享给别人,也不要粘贴到聊天里。
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": {
"webfetch": {
"command": "npx",
"args": [
"-y",
"getwebfetch-mcp"
],
"env": {
"UNSPLASH_ACCESS_KEY": "your-value",
"PEXELS_API_KEY": "your-value",
"PIXABAY_API_KEY": "your-value",
"BRAVE_API_KEY": "your-value",
"SERPAPI_KEY": "your-value"
}
}
}
}在终端中运行这条命令
claude mcp add webfetch --env UNSPLASH_ACCESS_KEY=your-value --env PEXELS_API_KEY=your-value --env PIXABAY_API_KEY=your-value --env BRAVE_API_KEY=your-value --env SERPAPI_KEY=your-value -- npx -y getwebfetch-mcp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": {
"webfetch": {
"command": "npx",
"args": [
"-y",
"getwebfetch-mcp"
],
"env": {
"UNSPLASH_ACCESS_KEY": "your-value",
"PEXELS_API_KEY": "your-value",
"PIXABAY_API_KEY": "your-value",
"BRAVE_API_KEY": "your-value",
"SERPAPI_KEY": "your-value"
}
}
}
}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": {
"webfetch": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"getwebfetch-mcp"
],
"env": {
"UNSPLASH_ACCESS_KEY": "your-value",
"PEXELS_API_KEY": "your-value",
"PIXABAY_API_KEY": "your-value",
"BRAVE_API_KEY": "your-value",
"SERPAPI_KEY": "your-value"
}
}
}
}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6日
注册名称已通过 GitHub 账户 ashlrai 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。
你需要填写:UNSPLASH_ACCESS_KEY, PEXELS_API_KEY, PIXABAY_API_KEY, BRAVE_API_KEY, SERPAPI_KEY
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(0.1.5)于6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。
License-first image search and download for any MCP-speaking agent.
Every result ships with a structured license tag (CC0, CC_BY, CC_BY_SA, platform-license tags, EDITORIAL_LICENSED, ...), a confidence score, and a ready-to-render attribution line. Covers 25 providers, with 19 in the default set — no API key needed for Wikimedia Commons, Openverse, iTunes, MusicBrainz CAA, NASA, The Met, Library of Congress, Internet Archive, Wellcome Collection, and Burst.
Most image-search tools give you URLs. webfetch gives you URLs plus the legal metadata needed to classify them as open/commercial reusable, platform-license, editorial-only, or exploration-only. The MCP layer exposes seven composable tools — search, specialize by artist/album, download with a hash, resolve an arbitrary URL's license, reverse-image-search, and triage a source page — so an agent can go from prompt to attributed asset in a single conversation turn.
attributionLine string, ready for a tooltip or credits footercurl -fsSL https://raw.githubusercontent.com/ashlrai/webfetch/main/install/install.sh | bash
Installs bun if missing, builds the server, symlinks webfetch onto $PATH, and merges the MCP block into ~/.claude/settings.json.
Add to ~/.claude/settings.json:
{
"mcpServers": { "webfetch": { "command": "bun", "args": ["run", "/path/to/webfetch/packages/mcp/src/index.ts"] } } }
Add to ~/.cursor/mcp.json:
{
"mcpServers": { "webfetch": { "command": "bun", "args": ["run", "/path/to/webfetch/packages/mcp/src/index.ts"] } } }
Paste into Cline's MCP settings JSON (VSCode sidebar → Cline → MCP Servers):
{
"mcpServers": { "webfetch": { "command": "bun", "args": ["run", "/path/to/webfetch/packages/mcp/src/index.ts"], "disabled": false, "autoApprove": ["search_images", "search_artist_images", "search_album_cover", "probe_page"] } } }
Replace /path/to/webfetch with your clone path (default after installer: ~/.webfetch/repo).
| Tool | Description | Key params |
|---|---|---|
search_images | Federated search across 25 providers. Returns the top candidates with license + attribution. Does not download. For large batches, use CLI handoff: webfetch batch --jsonl. | query, providers[], licensePolicy, minWidth, minHeight |
search_artist_images | Specialized artist image search with kind-aware provider routing: portrait, album, logo, performing. | artist, kind, providers[] |
search_album_cover | Canonical album artwork via MusicBrainz CAA + iTunes + Spotify. Results are EDITORIAL_LICENSED. | artist, album |
download_image | Download a URL to local disk cache. 20 MB cap, content-type guard, SHA-256 hash, host blocklist enforced. | url, maxBytes, cacheDir |
fetch_with_license | Resolve license for any URL via host heuristics + page metadata ( , dc.rights, OG tags). | url, probe (also download bytes) |
find_similar | Reverse-image-search a reference URL. Requires SERPAPI_KEY. Returns leads — verify before shipping. | url, providers[] |
probe_page | Enumerate every on a webpage with inferred dimensions and per-image heuristic license. Respects robots.txt. | url, respectRobots |
Results carry one of: CC0 · PUBLIC_DOMAIN · CC_BY · CC_BY_SA · UNSPLASH_LICENSE · PEXELS_LICENSE · PIXABAY_LICENSE · EDITORIAL_LICENSED · PRESS_KIT_ALLOWLIST · UNKNOWN (rejected by default). Use open-only when platform/editorial licenses are not acceptable.
Migration note: Unsplash, Pexels, and Pixabay are no longer reported as CC0. Agents should treat UNSPLASH_LICENSE, PEXELS_LICENSE, and PIXABAY_LICENSE as platform-license results, not Creative Commons/public domain results.
MCP tools are meant for interactive agent calls. For hundreds or thousands of queries, have the agent hand off to the CLI:
webfetch batch --file queries.txt --jsonl --continue-on-error --candidates 3
Input lines are query or query provider-a,provider-b. Each output line contains index, query, status, candidateCount, candidates, top, downloads, and provider diagnostics.
| Provider | Env var | Free tier? |
|---|---|---|
| Wikimedia Commons | — | Yes (no key) |
| Openverse | — | Yes (no key) |
| Unsplash | UNSPLASH_ACCESS_KEY | Demo key rate-limited |
| Pexels | PEXELS_API_KEY | Free tier available |
| Pixabay | PIXABAY_API_KEY | Free tier available |
| Spotify | SPOTIFY_CLIENT_ID + SPOTIFY_CLIENT_SECRET | Dev app required |
| Brave Search | BRAVE_API_KEY | Free tier available |
| SerpApi | SERPAPI_KEY | Required for find_similar |
| Bright Data managed browser | BRIGHTDATA_API_TOKEN | Optional server/cloud fallback |
MusicBrainz CAA and iTunes require no key. Missing keys cause the provider to skip gracefully — other providers still respond.
Try these in Claude Code or Cursor after installing:
Find me a CC-licensed portrait of Miles Davis I can use on an album liner page.
Search for album artwork for "Kind of Blue" by Miles Davis and show me the attribution line.
Probe the page at https://commons.wikimedia.org/wiki/File:Miles_Davis.jpg and tell me if I can use it commercially.