sarman/dgx-spark-playbooks
1
0
Fork 0
mirror of https://github.com/NVIDIA/dgx-spark-playbooks.git synced 2026-06-18 12:32:23 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
6a749bdcb0
dgx-spark-playbooks/nvidia/openclaw/README.md
GitLab CI b8cc262bed chore: Regenerate all playbooks
2026-06-13 02:56:21 +00:00

12 KiB
Raw Blame History

OpenClaw 🦞

Run OpenClaw locally on DGX Spark with a vLLM-served local model

Table of Contents

Overview

Basic idea

OpenClaw (formerly Clawdbot & Moltbot) is a local-first AI agent that runs on your machine. It combines multiple capabilities into a single assistant: it remembers conversations, adapts to your usage, runs continuously, uses context from your files and apps, and can be extended with community skills.

Running OpenClaw and its LLMs fully on your DGX Spark keeps your data private and avoids ongoing cloud API costs. DGX Spark is well suited for this: it runs Linux, is designed to stay on, and has 128GB memory, so you can run large local models for better accuracy and more capable behavior.

What you'll accomplish

You will have OpenClaw installed on your DGX Spark and connected to a local LLM served by vLLM (the agent-ready nvidia/Qwen3.6-35B-A3B-NVFP4 recipe). You can use the OpenClaw web UI to chat with your agent, and optionally connect communication channels and skills. The agent and models run entirely on your Spark—no data leaves your machine unless you add cloud or external integrations.

Popular use cases

  • Personal secretary: With access to your inbox, calendar, and files, OpenClaw can help manage your schedule, draft replies, send reminders, and find meeting slots.
  • Proactive project management: Check project status over email or messaging, send status updates, and follow up or send reminders.
  • Research agent: Combine web search and your local files to produce reports with personalized context.
  • Install helper: Search for apps/libraries, run installations, and debug errors using terminal access (larger models recommended).

What to know before starting

  • Basic use of the Linux terminal and a text editor
  • Optional: familiarity with Docker and vLLM if you plan to use a local model
  • Awareness of the security considerations below

Important: security and risks

AI agents can introduce real risks. Read OpenClaw’s guidance: OpenClaw Gateway Security.

Main risks:

  1. Data exposure: Personal information or files may be leaked or stolen.
  2. Malicious code: The agent or connected tools may expose you to malware or attacks.

You cannot eliminate all risk; proceed at your own risk. Critical security measures:

  • STRONGLY RECOMMENDED: Run OpenClaw on a dedicated or isolated system (e.g., a clean DGX Spark or VM) and only copy in the data the agent needs. Do not run this on your primary workstation with sensitive data.
  • Use dedicated accounts for the agent instead of your main accounts; grant only the minimum access it needs.
  • Enable only skills you trust, preferably those vetted by the community. Skills that provide terminal or file system access increase risk significantly.
  • CRITICAL: Ensure the OpenClaw web UI and any messaging channels are never exposed to the public internet without strong authentication. Use SSH tunneling or VPN if accessing remotely.
  • Where possible, limit internet access for the agent using firewall rules or network isolation.
  • Monitor activity: Regularly review logs and commands executed by the agent.

Prerequisites

  • DGX Spark running Linux, connected to your network
  • Terminal (SSH or local) access to the Spark
  • For local LLMs: enough GPU memory for your chosen model (see Instructions for size guidance; DGX Spark’s 128GB supports large models)

Time and risk

  • Duration: About 30 minutes for install and first-time model setup; model download time depends on size and network (the NVFP4 checkpoint is downloaded once and cached for later launches).
  • Risk level: Medium to High—the agent has access to whatever files, tools, and channels you configure. Risk increases significantly if you enable terminal/command execution skills or connect external accounts. Without proper isolation, this setup could expose sensitive data or allow code execution. Always follow the security measures above.
  • Rollback: You can stop the OpenClaw gateway and uninstall via the same install script or by removing its directory; stop the vLLM container separately (docker rm/docker rmi) if desired.
  • Last Updated: 06/12/2026
    • Switch local inference backend to vLLM (agent-ready Qwen3.6 35B recipe)
    • First Publication

Instructions

Caution

Before proceeding, review the security risks in the Overview tab. OpenClaw is an AI agent that can access your files, execute commands, and connect to external services. Data exposure and malicious code execution are real risks. Strongly recommended: Run OpenClaw on an isolated system or VM, use dedicated accounts (not your main accounts), and never expose the dashboard to the public internet without authentication.

Step 1. Install OpenClaw on your DGX Spark

On your DGX Spark, open a terminal and run the official install script. This installs OpenClaw and its dependencies on your Linux system.

curl -fsSL https://openclaw.ai/install.sh | bash

After dependencies are downloaded, OpenClaw will show a security warning. Read the risks; if you accept them, use the arrow keys to select Yes and press Enter.

Step 2. Complete the OpenClaw onboarding

Work through the prompts as follows.

  1. Quickstart vs Manual: Choose Quickstart.

  2. Model provider: To use a local model (recommended for DGX Spark), go to the bottom of the list and select Skip for now—you’ll configure the model later. To use a cloud model instead, pick a provider and follow its instructions.

  3. Filtering models by provider: Select All Providers. On the next prompt for the default model, choose Keep Current.

  4. Communication channel: You can connect a channel (e.g., messaging) to use the bot when away from the machine, or select Skip for Now and configure it later.

  5. Skills: We recommend selecting No for now. You can add skills later from the web UI or Clawhub after you’ve tested the basics.

  6. Homebrew: If you are prompted to install Homebrew, select No—Homebrew is for macOS only and is not needed on Linux.

  7. Hooks: We recommend selecting all three for a better experience. Note that this may log data locally; enable only if you’re comfortable with that.

  8. Dashboard URL: The terminal will print a URL for the OpenClaw dashboard. Save this URL (and any access token shown)—you’ll need it to open the web UI.

  9. Finish: Select Yes on the final prompt to complete installation.

You can now open the OpenClaw dashboard in a browser using the URL and token from the installer.

Step 3. Serve the model with vLLM on your DGX Spark

OpenClaw will connect to a local, OpenAI-compatible endpoint served by vLLM. This playbook uses the agent-ready nvidia/Qwen3.6-35B-A3B-NVFP4 recipe — the same one documented in the vLLM playbook's Run Agent Ready Qwen3.6 35B Model with vLLM tab. The NVFP4 quantization and speculative decoding give strong tool-calling and reasoning quality while leaving headroom on DGX Spark's 128GB unified memory.

In a separate terminal on your DGX Spark, follow the vLLM playbook's Run Agent Ready Qwen3.6 35B Model with vLLM tab to launch the server. Run it on its own terminal so the gateway and the model server can run side by side. That tab serves nvidia/Qwen3.6-35B-A3B-NVFP4 on an OpenAI-compatible API at http://localhost:8000/v1.

Once the server reports Application startup complete, verify it from another terminal before continuing:

curl http://localhost:8000/v1/models

You should see nvidia/Qwen3.6-35B-A3B-NVFP4 in the returned list.

Step 4. Configure OpenClaw to use the vLLM server

  1. Open the OpenClaw config file in your preferred editor (e.g. nano, vim, or a graphical editor). The config path is:

    ~/.openclaw/openclaw.json

    Example with nano:

    nano ~/.openclaw/openclaw.json

  2. Add or update the models section so it includes the vLLM provider pointing at the endpoint from Step 3. vLLM does not require an API key, so any non-empty placeholder works:

"models": {
  "mode": "merge",
  "providers": {
    "vllm": {
      "baseUrl": "http://localhost:8000/v1",
      "apiKey": "vllm",
      "api": "openai-responses",
      "models": [
        {
          "id": "nvidia/Qwen3.6-35B-A3B-NVFP4",
          "name": "nvidia/Qwen3.6-35B-A3B-NVFP4",
          "reasoning": true,
          "input": ["text"],
          "cost": {
            "input": 0,
            "output": 0,
            "cacheRead": 0,
            "cacheWrite": 0
          },
          "contextWindow": 262144,
          "maxTokens": 8192
        }
      ]
    }
  }
}

The id and name must match the model handle served by vLLM (nvidia/Qwen3.6-35B-A3B-NVFP4). contextWindow matches the --max-model-len from Step 3.

Note

If OpenClaw reports an unsupported-endpoint error against the Responses API, change "api": "openai-responses" to the OpenAI chat-completions variant for your OpenClaw version — vLLM always exposes /v1/chat/completions.

  1. If the OpenClaw gateway is already running, restart it so it reloads ~/.openclaw/openclaw.json and picks up the new provider.

Step 5. Verify the setup

  1. In a browser, open the OpenClaw dashboard URL (and use the access token if required).
  2. Start a new conversation and send a short message.
  3. If you get a reply from the agent, the setup is working.

You can also ask OpenClaw which model it’s using. In the gateway chat UI you can switch models by typing: /model MODEL_NAME (e.g. /model nvidia/Qwen3.6-35B-A3B-NVFP4).

Step 6. Optional: add skills and learn more

  • Skills add capabilities but also risk; only enable skills you trust (e.g., community-vetted ones). To add a skill:

    • Ask OpenClaw to configure a skill, or
    • Use the sidebar in the web UI to enable skills, or
    • Browse Clawhub for community skills.

  • For more usage and configuration details, see the OpenClaw documentation.

Troubleshooting

Symptom Cause Fix
OpenClaw dashboard URL not loading Gateway not running or wrong host/port Restart the OpenClaw gateway so it reloads ~/.openclaw/openclaw.json. Verify: Check that the gateway process is running with pgrep -f openclaw or ps aux | grep openclaw. Find URL/token: Check the original installer output (scroll up in your terminal) or look in gateway logs (typically ~/.openclaw/logs/) for the dashboard URL and access token
"Connection refused" to model (e.g. localhost:8000) vLLM server not running, still loading, or wrong port Confirm the vLLM container is up and finished loading (curl http://localhost:8000/v1/models lists the model) and that baseUrl in openclaw.json is http://localhost:8000/v1
OpenClaw says no model available Provider not configured or model handle mismatch Add the vllm provider to ~/.openclaw/openclaw.json and ensure id/name exactly match the served handle (nvidia/Qwen3.6-35B-A3B-NVFP4)
Out-of-memory or very slow inference on DGX Spark Model too large for available GPU memory or other GPU workloads Lower --gpu-memory-utilization or --max-model-len when launching vLLM, free GPU memory (close other apps), or check usage with nvidia-smi
Install script fails or dependencies missing Missing system packages on Linux Install curl and any required build tools; see OpenClaw documentation for current requirements
Config changes not applied Gateway not reloaded Restart the OpenClaw gateway so it reloads ~/.openclaw/openclaw.json