Feature Request: Use VS Code's Existing Kernel Picker for Python Injection

[devpkr1]

It would be great if Python Injection could reuse the same kernel picker that VS Code provides for Jupyter notebooks, instead of requiring users to open a notebook first.

Current Workflow

To use Python Injection, users typically need to open a Jupyter Notebook so that a Python kernel is available.

This adds an extra step when all that's needed is access to an existing Python environment.

Proposed Solution

Allow Python Injection to use the same "Select Another Kernel" dialog that VS Code already provides.

For example, users could choose from:

• Python Environments (Conda, venv, etc.)
• Jupyter Kernels
• Existing Jupyter Server
• Google Colab
• Existing JupyterHub Server

(See attached screenshot.)

Once a kernel is selected, Python Injection would execute code using that kernel, without requiring an open notebook.

Benefits

• No need to open a Jupyter Notebook just to run Python Injection.
• Reuse existing Conda and virtual environments.
• Support remote kernels (Jupyter Server, JupyterHub, Colab) if available through VS Code Extension.
• Consistent experience with the VS Code Jupyter extension.
• Faster and more streamlined workflow.

Screenshot

The feature would ideally use the same kernel picker that VS Code displays for notebooks:

Additional Notes

I'm not sure whether the VS Code Jupyter API exposes this kernel picker for extensions, but if it does, integrating it would make Python Injection much more convenient.

[mikeburgh]

Just checked it out, not possible at the moment as the native picker can be opened for a notebook, but it only shows controllers registered for that notebook type. Jupyter registers its kernels for Jupyter notebooks and interactive windows, not DBCode notebooks.

Its extension API only gives me access to kernels that have already started for an open notebook, so until they change that, I cannot get it any further.