Kernels

Run your notebook and flow code against Jupyter kernels — local or remote — from one place. Register kernel endpoints, execute code over Jupyter, and keep track of which kernel is running which file.

What you can do

• Register Jupyter kernel endpoints (local or remote) and see them in the Kernels sidebar, each with its label, CPU/GPU type, and address.
• Set a default kernel, test an endpoint's connection, and edit or remove it.
• Run code from notebooks, flows, and agents against a connected kernel.
• See active sessions — which kernel is attached to which file.
• Connect to token-protected endpoints; the token is stored securely, not in the endpoint list.

Getting started

Open the Kernels view from the activity bar, then use ＋ Add Kernel:

• Add Runtime — connect a remote Jupyter endpoint (needs the Runtime Kernels extension).
• Start Local Jupyter — launch a managed local Jupyter server (needs the Local Jupyter extension).

To inspect an endpoint's packages and connection health, add the Kernel Explorer extension.

Requirements

• A remote Jupyter endpoint works in either the browser or the desktop app.
• Running a local Jupyter server, or local processes, requires the desktop app.