Figma Context MCP
Give your coding agent access to your Figma data. Implement designs in any framework in one-shot.
Description from the maker (original English)
Control CesiumJS 3D globes via MCP: camera, layers, entities, animation, and interaction.
These three are never merged into one score: opinions, measurements and popularity are different things.
We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.
Pick the AI app you use, then copy the settings.
You need
Which app do you use?
Pick your app above. We remember it, so next time the button says “Install in …”.
Open Claude Desktop → Settings → Developer → Edit Config, and paste this into claude_desktop_config.json
{
"mcpServers": {
"cesium-runtime": {
"command": "npx",
"args": [
"-y",
"cesium-mcp-runtime"
]
}
}
}Run this in your terminal
claude mcp add cesium-runtime -- npx -y cesium-mcp-runtimePaste into .cursor/mcp.json in your project, or ~/.cursor/mcp.json
{
"mcpServers": {
"cesium-runtime": {
"command": "npx",
"args": [
"-y",
"cesium-mcp-runtime"
]
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
Paste into .vscode/mcp.json in your project
{
"servers": {
"cesium-runtime": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"cesium-mcp-runtime"
]
}
}
}Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.
This one runs on your computer, and ChatGPT can only connect to servers on the internet. Pick another app.
Your choice stays on this device in a cookie; no account needed.
These settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.
Runs on your computer; the tool list appears in your AI app after you install it.
We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 7, 2026
The registry name was verified with the GitHub account gaopengbin, which owns the repo
We have not read this repo from GitHub yet
We have not read this repo from GitHub yet
This one runs on your computer; there is no internet endpoint
Runs on your computer and uses local settings instead
Runs on your computer; we can only read tool lists from internet endpoints
Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.
It passed 1 of 1 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.
The registry does not list a required key, though some servers ask you to sign in on first use.
It runs on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).
The GitHub repo last changed –, and the latest version (2.0.0) reached the registry yesterday.
Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need Node.js on your computer.
In the same category: Figma Context MCP, Claude Real Video, Figma MCP Server
Servers in the same category from other makers, sorted by stars.
Reading a card:★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi
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.
Figma’s official server gives AI your design data so it can turn designs into code or describe them accurately. Connects to Figma’s cloud.
Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.
Usage statistics are on by default, with a startup notice on stderr. Set CESIUM_TELEMETRY_DISABLED=1, DO_NOT_TRACK=1, or pass --no-telemetry to disable. Collected fields and privacy.
English | 中文
MCP Server that enables AI Agents to control a Cesium globe in real-time via the Model Context Protocol.
Release channel:
latestuses the stable MCP SDK v2 and supports both
MCP
2025-11-25and2026-07-28from the same stdio/HTTP entry.
If you only want to try the "AI + Cesium" feel,
examples/browser-agent is the recommended
starting point — zero backend, runs in three minutes. See
openCesiumMap serves an interactive Cesium Viewer through the standard ui://cesium-mcp/map-v10.html MCP Apps resource. The map occupies the primary area and uses the host's existing native chat through ui/message, with no duplicate composer or history. Layout and host theme/locale changes preserve the Viewer, selection and session. The app follows host semantic style variables and fonts, supports Chinese and English, and publishes camera, layers and selected-object context. Local preview controls simulate live host settings and record message delivery. OpenAI global/thread entrypoint and fullscreen hints are included.
The built HTML, embedded Worker graph and Bridge bundle ship in dist. A compatible host must support WebGL and Blob module workers in an isolated, stable UI origin. The embedded panel exchanges commands through app-only MCP tools and does not access a local WebSocket. CESIUM_APP_DOMAIN optionally declares a dedicated UI origin. External browser Viewers retain the existing WebSocket transport.
For a real local MCP Apps preview, run npm run app:preview -w packages/cesium-mcp-runtime from a built repository and open http://127.0.0.1:19311/. See the full guide. Client-specific ChatGPT installation is a separate verification step.
Enable the tiles toolset with CESIUM_TOOLSETS=view,entity,layer,interaction,tiles or enable_toolset. Use loadVectorTiles for MVT /{z}/{x}/{y} URLs or vector 3D Tiles, click a feature and read getSelectedTileFeature, then update its managed layer using updateLayerStyle with tileStyle.color and tileStyle.lineWidth. Read back getLayerSchema.metadata.tileStyle to verify the applied expressions.
Vector rendering and clamping require CesiumJS 1.147. Clamped line/polygon picking uses an experimental bounded adapter. For external data in an MCP Apps panel, configure exact allowed origins with CESIUM_MAP_DATA_ORIGINS. See the vector tiles guide for source formats, styling and limits.
AI Agent <--MCP stdio--> cesium-mcp-runtime <--WebSocket--> Browser (cesium-mcp-bridge)
AI Agent <--MCP HTTP---> cesium-mcp-runtime <--WebSocket--> Browser (cesium-mcp-bridge)
The runtime acts as a bridge between MCP-compatible AI clients (Claude Desktop, VS Code Copilot, Cursor, etc.) and a browser running CesiumJS. It translates MCP tool calls into WebSocket commands that cesium-mcp-bridge executes.
The npm package ships its browser Bridge bundle locally. The built-in Viewer at http://localhost:9100/ therefore does not depend on an unpkg fallback; CesiumJS itself is still loaded from the official Cesium CDN.
For ordinary MCP users this is a one-package setup: run cesium-mcp-runtime and use the built-in Viewer. Install cesium-mcp-bridge separately only when developing a custom CesiumJS page.
Two transport modes are supported:
| Transport | Use Case | Protocol |
|---|---|---|
| stdio (default) | Local AI clients (Claude Desktop, VS Code, Cursor) | Standard I/O |
| http | Remote/cloud MCP clients (Dify, custom backends) | Streamable HTTP |
# Stable channel (stdio mode, default)
npx -y cesium-mcp-runtime
npm install -g cesium-mcp-runtime cesium-mcp-runtime
Then open http://localhost:9100/ to use the built-in Viewer. No separate Bridge installation is required for this default path.
For remote/cloud MCP clients like Dify:
# Start in HTTP transport mode
npx -y cesium-mcp-runtime --transport http --port 3000
Environment variable alternative:
MCP_TRANSPORT=http MCP_HTTP_PORT=3000 npx -y cesium-mcp-runtime
In HTTP mode, all 62 Cesium command tools are enabled by default (no dynamic toolset discovery needed).
Runtime binds to 127.0.0.1 by default, including its HTTP and WebSocket ports. Local native MCP clients and pages served from localhost continue to work. Pages opened directly as file:// must instead be served by a local HTTP server.
| Variable | Purpose |
|---|---|
CESIUM_HOST | Bind address, default 127.0.0.1. Non-loopback addresses require CESIUM_AUTH_TOKEN. |
CESIUM_AUTH_TOKEN | Shared connection secret. When set, all APIs and WebSocket connections require authentication, including local clients. |
CESIUM_ALLOWED_HOSTS | Comma-separated exact hostnames without ports, in addition to loopback and the bind address. No wildcards. |
CESIUM_ALLOWED_ORIGINS | Comma-separated exact HTTPS page origins. Loopback HTTP/HTTPS origins are allowed by default. No wildcards. |
For remote access, use HTTPS/WSS through a reverse proxy and configure the public hostname and Viewer origin explicitly. For example, set CESIUM_HOST=0.0.0.0, CESIUM_ALLOWED_HOSTS=maps.example.com, CESIUM_ALLOWED_ORIGINS=https://maps.example.com, and a generated CESIUM_AUTH_TOKEN in the server environment. Preserve the public Host header and WebSocket upgrades in the proxy.
MCP HTTP and REST clients send Authorization: Bearer . The built-in Viewer asks for the connection token when authentication is enabled; it keeps the token in the page only. Custom browser bridges authenticate with WebSocket subprotocols:
const encoded = btoa(String.fromCharCode(...new TextEncoder().encode(token)))
.replace(/\+/g, '-').replace(/\//g, '_').replace(/=+$/, '') const ws = new WebSocket('wss://maps.example.com/?session=demo', [ 'cesium-mcp', cesium-token.${encoded}, ])
Do not put credentials in query strings. Reverse proxies must not log Authorization or Sec-WebSocket-Protocol headers. The Viewer HTML and Bridge script are public assets; protected APIs and WebSocket upgrades still require the token. Secondary local Runtime instances must use the same token to relay to an existing instance. Browser session IDs select a Viewer; they are not credentials.
{
"mcpServers": { "cesium": { "command": "npx", "args": ["-y", "cesium-mcp-runtime"], "env": { "CESIUM_WS_PORT": "9100", "DEFAULT_SESSION_ID": "default" } } } }
In .vscode/mcp.json:
{
"servers": { "cesium": { "command": "npx", "args": ["-y", "cesium-mcp-runtime"], "env": { "DEFAULT_SESSION_ID": "default" } } } }
In .cursor/mcp.json:
{
"mcpServers": { "cesium": { "command": "npx", "args": ["-y", "cesium-mcp-runtime"] } } }
Tools are organized into 12 toolsets. By default, 4 core toolsets are enabled (30 tools). Additional toolsets can be activated via environment variable or dynamically by the AI agent at runtime. Shared titles, behavior annotations, localized descriptions, defaults, input validation, advertised output schemas, and structured results are generated from the canonical JSON Schemas in cesium-mcp-contracts. Tool calls also retain text content for older clients.
| Toolset | Tools | Default | Description |
|---|---|---|---|
view | 8 | Yes | Camera view controls + viewpoint bookmarks + scene export |
entity | 10 | Yes | Core entity operations + batch, query & property inspection |
layer | 9 | Yes | Layer management (GeoJSON, schema, style, basemap) |
interaction | 3 | Yes | Screenshot, highlight & measurement |
camera | 4 | — | Advanced camera controls (orbit, lookAt) |
entity-ext | 7 | — | Extended entity types (box, cylinder, wall, etc.) |
animation | 8 | — | Animation system (waypoints, clock, tracking, lighting) |
tiles | 7 | — | 3D Tiles, Gaussian Splats, terrain, imagery services, CZML & KML |
trajectory | 1 | — | Trajectory playback |
heatmap | 1 | — | Heatmap visualization |
scene | 3 | — | Scene options, post-processing & Runtime-only Ion credentials |
geolocation | 1 | — | Geocoding — convert address/place name to coordinates (Nominatim/OSM) |
{
"mcpServers": { "cesium": { "command": "npx", "args": ["-y", "cesium-mcp-runtime"], "env": { "CESIUM_TOOLSETS": "all" } } } }
CESIUM_TOOLSETS value | Result |
|---|---|
| *(not set)* | Default 4 toolsets (30 tools + 2 meta-tools) |
view,entity,camera,animation | Only specified toolsets + 2 meta-tools |
all | All 62 command tools, no meta-tools |
When not in all mode, two meta-tools are always available so the AI can discover and activate additional capabilities on demand:
| Tool | Description |
|---|---|
list_toolsets | List all toolset groups with enabled status and tool names |
enable_toolset | Dynamically enable a toolset — new tools become immediately available |
| Tool | Description |
|---|---|
flyTo | Fly to coordinates (lon, lat, height, heading, pitch, roll, duration) |
setView | Set camera position instantly |
getView | Get current camera state |
zoomToExtent | Zoom to bounding box (west, south, east, north) |
| Tool | Description |
|---|---|
addMarker | Add a marker at coordinates |
addLabel | Add text labels to the map |
addModel | Add 3D model (glTF/GLB or Ion asset) |
addPolygon | Add polygon with styling |
addPolyline | Add polyline with styling |
updateEntity | Update entity properties |
removeEntity | Remove entity by ID |
| Tool | Description |
|---|---|
addGeoJsonLayer | Add GeoJSON with styling (choropleth, category, etc.) |
listLayers | List all layers |
removeLayer | Remove layer by ID |
setLayerVisibility | Toggle layer visibility |
updateLayerStyle | Change layer color/opacity/width |
setBasemap | Switch basemap |
| Tool | Description |
|---|---|
lookAtTransform | Orbit-style camera aim at a position (heading/pitch/range) |
startOrbit | Start orbiting the camera around current center |
stopOrbit | Stop orbit animation |
setCameraOptions | Configure camera controller (enable/disable rotation, zoom, tilt) |
| Tool | Description |
|---|---|
addBillboard | Add an image icon at a position |
addBox | Add a 3D box with dimensions and material |
| addCorridor | Add a corrido
Only the beginning is shown; read the rest in the repo