impeccable
The design vocabulary for your coding agent. Steer with polish and bolder, review with critique and audit, catch AI slop before it ships.
Figma to pixel-perfect code: deterministic HTML/React, then a render + pixel-diff verify loop.
教 AI 做这类工作的 Skill,可与本项一起使用
The design vocabulary for your coding agent. Steer with polish and bolder, review with critique and audit, catch AI slop before it ships.
Editorial diagram design for Claude Code, Codex, GitHub Copilot, Cursor, Factory Droid, Qoder, and Pi. 45 diagram types. Self-contained HTML + SVG. No shadows. No Mermaid slop.
Create beautiful slides on the web using a coding agent's frontend skills
列表按使用分(满分 100)排序,由以下部分组成
| 热度 | 4/80 | GitHub 星标或每周下载量:较大的计 60,较小的计 20 |
|---|---|---|
| 新鲜度 | 12/12 | 30 天内有发布或代码更新得满分 |
| 仍在响应 | 0/8 | 上次检查时云端端点有响应 |
| 使用分 | 16/100 |
质量: 基础检查通过 3/3 项(仅展示,不用于排序) · 上升势头: 暂无数据 (只用于“上升中”标签)
选择你使用的 AI 应用,然后复制设置。
通过 3/3 项检查 · 在你的电脑上运行
第一步:你用哪个 AI?
选择你用的应用,步骤会显示在下方。
Claude 暂时无法连接 figma-mcp
还没有通过检查的安装方式。请看作者的说明,或选择同类的其他服务器
ChatGPT 暂时无法连接 figma-mcp
还没有通过检查的安装方式。请看作者的说明,或选择同类的其他服务器
B · 设计、图片与音频 类中能连接 ChatGPT 的其他服务器
从同类中使用较多的里挑选,仅限用地址即可连接、无需密钥的
Claude Code 暂时无法连接 figma-mcp
还没有通过检查的安装方式。请看作者的说明,或选择同类的其他服务器
Cursor 暂时无法连接 figma-mcp
还没有通过检查的安装方式。请看作者的说明,或选择同类的其他服务器
VS Code 暂时无法连接 figma-mcp
还没有通过检查的安装方式。请看作者的说明,或选择同类的其他服务器
还没有通过检查的安装方式。请看作者的说明,或选择同类的其他服务器
FIGMA_TOKEN 机密 必填 envFIGMA_MCP_CACHE_DIR env说明为自动翻译,变量名保持原样;不确定时请查看原文。
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。
依据这个服务器的连接方式和各应用自己的文档,检查于 2026年11月10日。
在你的电脑上运行,适用于
GooseLM StudioMsty StudioCherry StudioVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatRaycast
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19日
注册名称已通过 GitHub 账户 Mehmoodqureshi 验证,该账户拥有此仓库
代码最后更新于13天前
使用 MIT
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
OSV.dev 中没有 @mehmoodqureshi/figma-mcp@0.2.1 的记录
只检查这个版本的包本身,不包括所有依赖 · 检查于 2026年11月15日
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/servers/mehmoodqureshi-figma-mcp)提问、分享技巧或报告无法使用。按“有帮助”数量排序。
还没有评价
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 3 项适用的基础检查中通过了 3 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:FIGMA_TOKEN
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13天前,最新版本(0.2.1)于2个月前进入注册中心。
同一分类中有:HyperFrames by HeyGen, Pascal 3D Editor, draw.io, Figma MCP Server, Remocn: Remotion MCP server
同一分类、其他作者、命令已通过检查。官方的排在前面,再按星标排序。
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
Official HyperFrames by HeyGen: the best AI video editor in chat. Captions, motion graphics, MP4.
Create, inspect, validate, and save editable 3D building scenes with Pascal's MCP server.
Create draw.io diagrams in chat from Mermaid or XML, shown live and interactive. 10,000+ shapes.
The Figma MCP server brings Figma design context directly into your AI workflow.
Hosted MCP for Remotion videos: search Remocn components, read docs, plan a video, preview it.
| 本项 figma-mcp | HyperFrames by HeyGen | Pascal 3D Editor | draw.io | Figma MCP Server | Remocn: Remotion MCP server | |
|---|---|---|---|---|---|---|
| 运行位置 | 需要安装在电脑上 | 在云端运行,无需安装 | 云端或安装在电脑上都可以 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 |
| 需要密钥 | 是 | 否 | 否 | 否 | 否 | 否 |
| 可在 Claude 中按 + 添加 | 否 | 是 | 否 | 否 | 是 | 否 |
| 基础检查 | 3/3 | 4/4 | 5/5 | 4/4 | 4/4 | 4/4 |
| 许可证 | MIT | Apache-2.0 | MIT | Apache-2.0 | – | MIT |
| 价格 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 |
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。
Figma → code that is actually checked against the design. The pixel-diff verifier for Figma-to-code: it renders the output and diffs it against the frame until it converges — and it can verify *your own running app* the same way, so it works alongside Figma's official MCP server or Framelink rather than replacing them.
Most Figma-to-code tools stop at the moment of generation: they hand you JSX and walk away. Nobody ever renders the output and compares it to the frame, so the result stalls at "90% there" and you spend the afternoon nudging padding by eye.
This MCP server closes that loop. It converts a frame to self-contained HTML deterministically (no LLM, no hallucinated layout), then renders your code in Chromium, pixel-diffs it against the Figma reference, measures every element's box, and tells the agent exactly what is missing or misplaced. The agent edits, calls verify again, and repeats until the diff is at the anti-aliasing floor.
Figma reference First pass — 5.18% off Pixel diff Converged — 0.00%
Those are real artifacts from npm run example, which runs the loop end to end with no API key. The generate callback in that demo is a *scripted* stand-in for an LLM, so it shows the loop mechanics honestly — the measuring is real, the "model" is not.
A Figma node tree tells you what the designer *declared*. It does not tell you what a browser will *do* with your CSS. Those diverge constantly — a flex gap that collapses, a font that falls back, an absolute child that escapes its parent. The only way to know is to render it and look.
figma_verify gives the agent three independent signals per pass:
data-ir-id, measured in the live DOM, and matched against its Figma box. This is what turns "the bottom-left looks off" into "the CTA button is missing" or "the price label is 40px too low".elementsFromPoint over a sample grid gives the rendered order wherever two elements actually overlap. This catches the class of bug the other two signals are blind to by construction: elements present, in exactly the right box, stacked the wrong way round — or cut off by an ancestor's overflow where the design lets them overhang. Neither moves a box, and on a tall frame neither moves the diff ratio much, but a decorative overlay drawn over a photo instead of under it is the first thing a person sees. Reported as WRONG STACKING / OVER-CLIPPED, and reported even when the pixel diff has converged.| Tool | What it does | Network |
|---|---|---|
figma_convert | Fetch a frame → HTML + ir.json + images as files in assets/ (or data URIs with inlineAssets: true) + reference.png + design tokens (tokens.json in W3C DTCG format, tokens.css, a Tailwind v4 tailwind.theme.css). Optional react: true (component + the components/ it imports), next: true (a runnable App Router project) and tailwind: true (React/Next.js output as Tailwind classes), plus a responsive variant. | Figma API (cached) |
figma_verify | Render HTML in Chromium, pixel-diff vs the reference, IoU-check every element, check paint order and clipping, return targeted fix instructions. Pass url (and optionally selector) to verify a running page instead — your own app at http://localhost:3000, in any framework — against the frame. | none (or your dev server) |
figma_inspect | Print the IR as an indented outline (role, box, layout, text, tokens), filterable — read the structure without dumping raw Figma JSON into context. | none |
responsive: true reads Figma's sizing intent — FILL becomes flex/100%, HUG becomes fit-content, FIXED keeps its px. That only exists where the designer used Auto Layout.
A frame with no Auto Layout on the root is a canvas: every child is position:absolute at a coordinate on a 1440px artboard. There is nothing to relax, so making the root width:100% does not reflow it — the children stay pinned at their canvas coordinates and the right-hand side disappears under overflow:hidden. That reads exactly like a page breaking when you zoom.
For those frames the converter keeps the exact canvas and scales it to the viewport instead, so the design stays intact at every width and zoom level (down to 0.5x, after which the page scrolls). figma_convert tells you which strategy it used. If you want real reflow rather than proportional scaling, add Auto Layout in Figma — that is the signal the converter needs.
Every tool returns file paths and numbers, never large blobs. A converted frame is often 100+ KB of HTML; pushing that through a tool result would burn the agent's context for nothing. The agent reads and edits the files directly.
Figma responses are cached per frame, so after the first figma_convert the whole loop runs offline and free — including when you are rate-limited.
You need a Figma personal access token: Figma → Settings → Security → Personal access tokens, scope File content: read.
export FIGMA_TOKEN=figd_REPLACE_WITH_YOUR_TOKEN
Add it to Claude Code:
claude mcp add figma --env FIGMA_TOKEN=$FIGMA_TOKEN -- npx -y figma-codekit-mcp@latest
Or for any MCP host that reads a JSON config:
{
"mcpServers": { "figma": { "command": "npx", "args": ["-y", "figma-codekit-mcp@latest"], "env": { "FIGMA_TOKEN": "figd_REPLACE_WITH_YOUR_TOKEN" } } } }
Then, in the agent: copy a frame link out of Figma (right-click the frame → Copy link to selection) and say *"convert this frame and verify it until it converges."*
FIGMA_TOKEN=$FIGMA_TOKEN npx -y figma-codekit-mcp@latest --check
It checks Node, whether a token was found (and where, never its value), whether Figma accepts it, and whether the Chromium figma_verify renders in is installed, then prints READY or what to fix. Exit code 0 means all three tools will work. If Chromium is missing (a skipped download, or CI), install it with npx -y figma-codekit-mcp@latest --install-browser.
npx resolves to npx.cmd, which some MCP hosts cannot spawn directly. If the server fails to start, point at the shim explicitly:
{
"command": "cmd", "args": ["/c", "npx", "-y", "figma-codekit-mcp@latest"] }
figma_convert → generated.html, reference.png, ir.json (deterministic, no LLM)
↓ figma_verify → "NOT CONVERGED — 4.93% of pixels differ. Elements: 22/26 match — 1 missing, 3 misplaced. MISSING: button 'Get started' MISPLACED: h1 'Pricing plan' — 6px too high" ↓ agent edits generated.html ↓ figma_verify → "CONVERGED — 0.31% of pixels differ (threshold 2.00%)."
The refine loop runs through the calling agent, not through an LLM inside the server. That means no ANTHROPIC_API_KEY, no second model billing, and the agent keeps full context on what it already tried. A headless refine loop (src/refine/) still exists for library use.
The same pipeline has a one-page local web front end, split in two: a chat on the left, the generated app on the right. You send a frame link, it asks what to build — HTML, React or Next.js; exact or responsive — and then converts, renders and diffs in front of you. The right pane carries the running app, its source files, the verify numbers and the pixel diff.
npm run site # http://localhost:5173
It uses the same .figma-token and the same .figma-cache/, so a frame you have already converted re-runs offline in about a second. Useful for checking a frame converts cleanly before pointing an agent at it — and for showing someone what "verified against the design" means rather than describing it. Details in site/README.md.
| Variable | Purpose |
|---|---|
FIGMA_TOKEN | Required. Also read from a .figma-token file in your project directory. FIGMA_API_KEY is accepted as an alias. |
FIGMA_MCP_CACHE_DIR | Where converted frames are cached. Defaults to .figma-cache/ in the directory the server is started from. |
ANTHROPIC_API_KEY / GEMINI_API_KEY | Optional, headless refine loop only. The MCP server never calls an LLM. |
FIGMA_MCP_SKIP_BROWSER_DOWNLOAD=1 | Skip the Chromium download on install. figma_convert and figma_inspect still work; figma_verify will not. |
See .env.example. Add .figma-cache/ to your project's .gitignore — a cached frame keeps the original bitmaps and can be tens of megabytes.
FIGMA_FILE_TIMEOUT_MS (default 90000) caps how long figma_convert waits for Figma's whole-file download, which it falls back to when the per-node endpoint is rate-limited. When Figma will not answer, the tool says so within that time and, if the frame is already cached anywhere, builds from the cached copy and marks the result STALE CACHE instead of hanging.
figma_convert url=<frame link> → .figma-cache/<frame>/ (reference.png)
figma_verify dir= url=http://localhost:3000/pricing selector="#plans"
Build the page however you like — Figma's official MCP server, Framelink, by hand — then point figma_verify at it. It screenshots the page (or just the selected element) at the frame's size and pixel-diffs it against Figma. Element-box and layering checks need the converter's data-ir-id markers, so for a URL only the pixel diff runs.
Being straight about the edges, because they are Figma's, not bugs:
tokens.json / tokens.css / tailwind.theme.css, but names them after the value (--color-c8960c) because it cannot know intent. Rename them in your theme.figma_convert tells you when the limit lifts.threshold defaults to 2% for that reason — chasing 0 is chasing rendering noise.The verify loop is framework-agnostic and does not need the MCP layer:
import { verifyLoop } from 'figma-codekit-mcp';
import fs from 'node:fs';
const result = await verifyLoop(