NVIDIA USDCode MCP Server
Provides access to NVIDIA USDCode via the NVIDIA Integrate API for Isaac Sim scripting assistance, USD guidance, Python code snippets, and API usage help
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@NVIDIA USDCode MCP ServerHow do I create a simple robot arm in Isaac Sim using USD?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
nvidia-usdcode-mcp-server
A minimal MCP server that exposes a single tool backed by NVIDIA USDCode via the NVIDIA Integrate API. Great for Isaac Sim scripting tips, USD, Python snippets, and API usage.
Requirements
Node.js 18+
An NVIDIA API key in
NVIDIA_API_KEY, get it from here https://build.nvidia.com/nvidia/usdcode
You can place it in a local .env file (loaded automatically) or export it in your shell.
Related MCP server: Isaac Sim MCP Server
Quickstart
Dependencies are already vendored in this workspace. If you need to reinstall:
npm ciornpm install
Build
npm run build
Outputs to dist/.
Run (standalone)
npm start(after building)Or directly:
node --enable-source-maps dist/server.js
The server speaks MCP over stdio and is intended to be launched by an MCP-capable client.
Use with Claude Desktop
Add an entry to your ~/.claude/config.json:
{
"mcpServers": {
"usdcode": {
"command": "node",
"args": ["dist/server.js"],
"env": { "NVIDIA_API_KEY": "YOUR_KEY" }, // if you didn't add .env to this project
"disabled": false
}
}
}Use with OpenAI Codex
Add an entry to your ~/.codex/config.toml:
[mcp_servers.usdcode]
command = "node"
args = ["/absolute/path/to/mcp-usdcode/dist/server.js"]
env = { "NVIDIA_API_KEY" = "value" }Tool Info
name:
get_usdcode_helpdescription: Ask NVIDIA USDCode for help.
params:
question(string, required): Your prompt or question.temperature(number, optional, default 0.1)top_p(number, optional, default 1)max_tokens(integer, optional, default 1024)expert_type(string, optional, default "auto")
Returns a single text message with the model’s reply.
Available Tools
1 toolget_usdcode_helpC
Ask NVIDIA USDCode for help (Isaac Sim scripting, USD, Python/API tips).
Parameters: temperature (0-1, default 0.1), top_p (<=1, default 1), max_tokens (1-2048, default 1024), expert_type (auto|knowledge|code|helperfunction; default auto), stream (boolean; default false). Avoid changing temperature and top_p together.
| Name | Required | Description | Default |
|---|---|---|---|
| question | Yes | Your prompt or question | |
| temperature | No | Sampling temperature (0-1). Default: 0.1 | |
| top_p | No | Top-p nucleus sampling mass (<=1). Default: 1 | |
| max_tokens | No | Max tokens to generate (1-2048). Default: 1024 | |
| expert_type | No | Expert to use: auto, knowledge, code, or helperfunction. Default: auto | |
| stream | No | Stream partial deltas via SSE. Default: false |
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
1 tool update
v1.0.0- Changed
get_usdcode_help2 fields changed- added
Input schema / $schemaAdded value: +"http://json-schema.org/draft-07/schema#" - added
Input schema / additionalPropertiesAdded value: +false
1 tool update
- First observed
get_usdcode_help
TDQS
Scored across 1 tool
With only one tool, there is no possibility of ambiguity or overlap between tools. The single tool has a clearly defined purpose of providing help with NVIDIA USDCode topics, so an agent cannot misselect between multiple options.
Since there is only one tool, naming consistency is inherently perfect. The tool name 'get_usdcode_help' follows a clear verb_noun pattern, and there are no other tools to compare it against for inconsistency.
A single tool is too few for a server with a broad purpose like 'NVIDIA USDCode MCP Server', which suggests coverage of Isaac Sim scripting, USD, and Python/API tips. This minimal toolset feels thin and under-scoped, limiting the server's utility for diverse agent tasks.
The tool surface is severely incomplete for the implied domain. While 'get_usdcode_help' might assist with queries, there are obvious gaps such as tools for executing code, managing USD files, or interacting with Isaac Sim APIs. This will likely cause agent failures when trying to perform specific actions beyond general help.
Maintenance
Related MCP Connectors
Connect your AI assistant to Blender for 3D modeling, Geometry Nodes, materials, lighting, animation, rendering and export. Inspect real scenes and viewport images, refine the result and restore checkpoints. OAuth connects your chat to Blender Bridge Desktop on an online Windows 10/11 x64 computer running Blender 4.2+. The chat can run on another device. Requires an active Blender Bridge account; 7-day trial available. AI model access is separate.
Use AI models for chat, image, and video generation from Claude Code and other MCP hosts.
Build, edit and publish live 3D, AR and game scenes in ARClip from your AI assistant.
Run accessibility, contrast, and design token audits and file QA issues from your AI coding tool.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to read, inspect, and manipulate OpenUSD 3D scenes, including scene graph traversal, property querying, mesh export, and variant switching.10MIT
- AlicenseAqualityBmaintenanceEnables natural language control of NVIDIA Isaac Sim through the Model Context Protocol, allowing you to create robots, build scenes, run simulations, and debug physics from any MCP-compatible IDE.4234 PyPI69MIT
- FlicenseAqualityDmaintenanceEnables AI coding assistants to browse Unreal Python API documentation, search classes and functions, and execute Python code in Unreal Editor.8-
- AlicenseNot gradedqualityDmaintenanceEnables AI tools like Windsurf and Claude to control NVIDIA Isaac Sim and Isaac Lab through natural language, providing tools for scene inspection, prim management, physics simulation, and robot spawning.8Apache 2.0