NVIDIA NemoClaw CKG

Name: NVIDIA NemoClaw CKG
Author: Yarmoluk

Yarmoluk

开始使用源码仓库

NVIDIA NemoClaw knowledge graph — 55 nodes, F1 0.576 (+269% vs RAG), 11x fewer tokens. MCP-native.

接入配置

选择客户端，复制配置，再按项目文档补齐环境变量。

一键添加到 Cursor 一键添加到 VS Code

本地运行

claude mcp add nvidia-nemoclaw-ckg -- uvx ckg-nvidia-nemoclaw

~/.codex/config.toml

[mcp_servers.nvidia-nemoclaw-ckg]
command = "uvx"
args = ["ckg-nvidia-nemoclaw"]

# codex mcp add nvidia-nemoclaw-ckg -- uvx ckg-nvidia-nemoclaw

~/.cursor/mcp.json

{
  "mcpServers": {
    "nvidia-nemoclaw-ckg": {
      "command": "uvx",
      "args": [
        "ckg-nvidia-nemoclaw"
      ]
    }
  }
}

.vscode/mcp.json

{
  "servers": {
    "nvidia-nemoclaw-ckg": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "ckg-nvidia-nemoclaw"
      ]
    }
  }
}

~/.gemini/settings.json

{
  "mcpServers": {
    "nvidia-nemoclaw-ckg": {
      "command": "uvx",
      "args": [
        "ckg-nvidia-nemoclaw"
      ]
    }
  }
}

claude_desktop_config.json

{
  "mcpServers": {
    "nvidia-nemoclaw-ckg": {
      "command": "uvx",
      "args": [
        "ckg-nvidia-nemoclaw"
      ]
    }
  }
}

.trae/mcp.json

{
  "mcpServers": {
    "nvidia-nemoclaw-ckg": {
      "command": "uvx",
      "args": [
        "ckg-nvidia-nemoclaw"
      ]
    }
  }
}

远程托管

claude mcp add --transport http nvidia-nemoclaw-ckg https://ckg-nvidia-nemoclaw.onrender.com/mcp