I used an AGPL-licensed library for a data study, and am hoping to release and link to a Jupyter Notebook with my methodology. I was hoping to learn what license, if any, I would need to include with my notebook.

Here's my situation:

  1. The notebook imports a forked version of an AGPL tool (the fork only differs by three lines of code meant to catch an exception, which is currently an Open PR to the original repo.) This fork is in a repo separate from my notebook.
  2. The notebook relies heavily on methods from the AGPL fork, but does not copy or modify source code from it.
  3. The notebook will be published standalone, in its own/separate repo, with a comment linking to the forked repo above the relevant import statement in the notebook.

From what I understand, the AGPL fork automatically inherits the AGPL license from the original repo. But I then am unsure about what happens when I import this fork into a notebook and publish it.

If my notebook imports a fork of an AGPL library, but I don't make any changes or copy source code from the fork into my Jupyter notebook, can I release a notebook that imports/makes heavy use of this library without needing to license the notebook?

My end goal is to link to the notebook in a data study post on a company blog so readers can see my methodology. I just want to make sure I am correctly licensing my study/providing proper attribution for this AGPL tool.

  • 2
    Is the modified fork to the AGPL-licensed your own fork? And if not, did you ask that person to create this fork on your behalf? Jan 31, 2017 at 8:22
  • @PhilippeOmbredanne The fork is a personal fork I made after the author of the main library requested a PR to fix an issue I opened. The PR remains open so I had to use the forked library for the notebook to work. The notebook repo is under a different organization (the company that I work for.) To be safe I may just carry over the AGPLv3 license to the notebook.
    – Brendan
    Feb 1, 2017 at 9:03
  • A difficult thing is whether a notebook is code or data. It is clearly a bit of both. I need to think about it more. Feb 3, 2017 at 7:26
  • Code, data, or rather documentation?
    – gerrit
    Nov 9, 2021 at 13:37

1 Answer 1


I would consider a notebook as a combo of code and data.

As data, the notebook itself (assuming it does not include snippets of AGPL-licensed code) can be under whatever license you please.

As code, if executed it depends on your modified AGPL-licensed library that you make otherwise available.

Since you intent is to publish in some blog post, you are imho staying in the realm of data.

I would accompany the post with some comments explaining the dependency and its license if someone would want to reuse the code and its dependency for something else.

  • Isn't a notebook rather like documentation?
    – gerrit
    Nov 9, 2021 at 13:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.