danielballan / conda-kernels

Load conda environments as Jupyter kernels.

conda-kernels

Inspect conda environments to automatically generate Jupyter kernels.

As noted in the code, this is adapted from this gist by Phil Elson (@pelson).

To make a conda environment "discoverable" by Jupyter, run

conda install -n MY_ENV_NAME -c danieballan/kernelize

(The conda receipe for kernelize is included in this package, so you can build it yourself and replace danielballan with your anaconda channel.)

Then, with conda-kernels installed, launch jupyter like so:

jupyter notebook --NotebookApp.kernel_spec_manager_class=conda_kernels.CondaKernelSpecManager

This is designed to be used with the remotekernel project, and currently it does not work without it. Making it work as a standalone would be straightforward. Get in touch if you need this.