Create JupyterLab environment (conda):

mamba env create -n jlab --file jlab-environment.yml

Create scratchpad uv config (creates pyproject.toml, uv.lock, .venv):

uv init
uv add ipykernel sunpy hapiclient matplotlib pycdfpp
uv add --editable ../SwarmPAL ../VirES-Python-Client

Install the kernel (for the scratchpad environment) in userspace so that jupyter will find it:

uv run python -m ipykernel install --user --name scratchpad

(the config is located in ~/.local/share/jupyter/kernels/)

Recreate uv.lock:

uv lock

Recreate .venv:

uv sync

• Use the "Manage Python environments" tool to point to the jlab conda environment
• It doesn't see the same .local/share/jupyter path to get kernels (because it is in Flatpak?)
  • (you can observe this by running jupyter --paths from a terminal in the JupyterLab)
  • (nb_conda_kernels is still working and the conda kernels are there)
  • To fix, open the JupyterLab Desktop Settings, and set the environment variable: JUPYTER_PATH= "/home/ash/.local/share/jupyter"

• Get a key from, e.g. https://platform.openai.com/settings/organization/api-keys