vbookshelf/Juru-Lab-Agent-Sandbox-HYA

Juru Lab Desktop Agent Sandbox

Giving desktop agents shell access is dangerous. They can delete your personal files. But not giving desktop agents shell access means learners miss the opportunity to see the relentlessness, intelligence and self correcting flexibility of modern AI agents. What's needed is an isolated local workspace where learners can see the agent run wild.

Juru-Lab is JupyterLab combined with the Shuru MicroVM engine. This creates a sandboxed workspace where agents can use shell access to install packages, create and run code, and manipulate files.

In other words, when the agent reads and writes files or when it creates and runs code, it's doing all this inside a Linux sandbox. It can't access or even see the files in your home directory. If you want to give it access to any files or folders, you need to place these inside the agent's workspace folder.

This is a Hybrid App (HYA). The first install requires an internet connection. After that it can run offline.

• No VM setup needed. The Shuru microVM binary is hosted locally.
• Simple to use. No Docker complexity.
• Includes Jupyter notebooks with example agentic loops for Ollama and OpenRouter.
• Agent loop includes pre-defined tools
• Agent loop includes support for Skills.
• No framework abstractions. Readable Python code that's easy to understand, learn from and build on.
• Linux data science packages are available for the agent to self-install in the microVM.
• More Linux packages can be downloaded.
• Agent loop includes printed outputs so it's easy to see what the agent is doing.

How to Use

System Requirements:

• Operating System: MacOS
• Computer: Apple Silicon Mac (M1, M2, M3, etc.)
• RAM: 16 GB
• Free disk Space: 8.5 GB

1. Install Ollama
   https://ollama.com/download/mac

2. Download the qwen3.5:9b (6.6GB) model from Ollama
   Paste this in the terminal:

ollama pull qwen3.5:9b

4. Download the Juru-Lab-Agent-Sandbox-v1.0-HYA.zip folder and unzip.
5. Right click on the Juru-Lab-Agent-Sandbox-v1.0-HYA.zip folder and select: New Terminal at Folder
6. MacOS often quarantines downloaded files. To make the launch script executable, paste this command into the terminal and press Enter:

cat start-mac-app.command > temp && mv temp start-mac-app.command && chmod +x start-mac-app.command

You only need to do this once, during the setup.

6. Open the Juru-Lab-Agent-Sandbox-v1.0-HYA.zip folder and double-click this file: start-mac-app.command
   If a MacOS security popup appears, click: "Allow"

7. Python data science packages will be downloaded using uv. These packages are in addition to the already included Linux data science packages that the agent can access.

8. JupyterLab will then open. This may take a few seconds. Please be patient if you don't see anything happening. During normal use the app will open much faster.

9. Open the folder named: Jupyter-Notebooks

10. Review the notebook named: 1-Quick-Start-Guide.ipynb

References

• Shuru microVM sandbox
  https://github.com/superhq-ai/shuru

Revision History

Version 1.0
21-May-2026
Prototype. Released for testing.