Jupyter Kernel Proxy

This Jupyter kernel serves as a proxy to any already running kernel. It is meant to allow Jupyter Notebook to use kernels it cannot start itself, such as kernel embedded in another applications.

Jupyter Kernel Proxy was initially created as a way to use Jupyter Notebook using IPyIDA, but should work with any other embedded or remote kernel.

Install

Jupyter Kernel Proxy is on PyPi and can be installed using

pip install jupyter-kernel-proxy

The kernelspec must also be installed using:

python -m jupyter_kernel_proxy install

Usage

In Jupyter Notebook, start a new notebook and choose "Existing session" as kernel type.

When launched, the kernel proxy will try to connect to the last started kernel. This is similar to what jupyter console --existing would do.

The kernel implements a magic %proxy command to allow connecting to other running kernels. Only two subcommands are implemented: list and connect.

Here is an example usage:

In  [ ]: %proxy list
           kernel-382cad4b-2286-4ea0-bcb7-fcc30039ac78.json (proxy)
         * kernel-55159.json (no name)
           kernel-55151d00-21e4-4ea9-8dfd-b0842e27cba1.json (no name)
           kernel-50889.json (no name)
In  [1]: a = 5
In  [ ]: %proxy connect 55151d00-21e4-4ea9-8dfd-b0842e27cba1
         Connecting to kernel-55151d00-21e4-4ea9-8dfd-b0842e27cba1.json
In  [1]: a = 4
In  [2]: a
Out [2]: 4
In  [ ]: %proxy connect kernel-55159.json
         Connecting to kernel-55159.json
In  [2]: a
Out [2]: 5