Work in VS Code

Most developers write their code in VS Code. This page shows how to use it with a stardelt platform: your notebooks and Python run on the platform, as you, next to the data, while you keep editing in VS Code with its debugger, Git integration and extensions.

What works today

You want to	How
Run notebooks and Python scripts against the warehouse	Connect VS Code to your JupyterHub server (below). The kernel runs on the platform; queries go to Trino under your own identity
Keep notebooks and code in Git	Use VS Code's Git support on your local checkout; the kernel only needs the platform
Edit pipeline and platform YAML	Any YAML editor; schema validation is not available yet

Not available yet: a SQL client with platform sign-in, VS Code in the browser, and schema-validated pipeline files. See What is planned.

Connect VS Code to your notebook server

You need: VS Code, the Python and Jupyter extensions (ms-python.python, ms-toolsai.jupyter), and a platform where JupyterHub is installed. Optionally install the JupyterHub extension (ms-toolsai.jupyter-hub); it lets you pick the hub instead of pasting a URL.

1. Start your server. Open https://jupyter.<your-domain> in a browser, sign in, and let the hub start your server. (The kernel only runs while your server runs.)

2. Create an API token. Open https://jupyter.<your-domain>/hub/token, create a token with a note such as "vscode", leave the permissions empty, and copy it. It is shown once.

3. Connect. In VS Code open or create a notebook (.ipynb), click Select Kernel, choose Select Another Kernel → Existing Jupyter Server → Enter the URL of the running Jupyter server, and paste

   https://jupyter.<your-domain>/user/<your-username>/?token=<your-token>

   Pick the Python kernel. The server name appears in the top right.

4. Check it runs on the platform. Run !hostname in a cell: it prints the pod name of your server, not your laptop.

Query the warehouse

Your server starts with everything needed to query as you: your sign-in token (TRINO_TOKEN), the Trino address (TRINO_URL), the platform's certificate bundle and the Trino client. The notebook warehouse-quickstart.ipynb is copied into your home directory on the first start; open it in VS Code (through the connected server's file browser in a browser tab, or copy its first code cell), run it, and reuse its sql() helper in your own notebooks. Its first query, SELECT current_user, prints your username, not a shared account.

The token belongs to your session. When it expires, restart your server from the hub to get a fresh one.

Troubleshooting

Symptom	Cause and fix
"Failed to connect" or timeout	Your server is not running: start it from https://jupyter.<your-domain>/hub/home, then retry
403 or "invalid token"	The token was copied with a trailing space, or has been deleted: create a new one at /hub/token
The kernel list is empty	The URL must end in /user/<username>/ (with the token query) and not point at the hub root
KeyError: 'TRINO_TOKEN'	You started the kernel outside the platform's server (a local kernel). Select the remote kernel again
Queries fail with 401 after hours	Your sign-in token expired: restart your server from the hub

What is planned

• A stardelt VS Code extension: sign in with your platform account, browse the catalog, run SQL and see results, start pipelines.
• JSON schemas for the platform's resources, so pipeline files get completion and validation in the YAML editor.
• VS Code in the browser as a workspace option next to JupyterLab.