Our GitHub: https://github.com/mljar/mercury

Sharing Python notebooks is challenging. You can't send notebook directly to non-technical stakeholders. You need to copy-paste results/charts into Word/PowerPoint or rewrite notebook to web framework. Mercury converts notebook to web app. User can execute cells but can't edit them.

Mercury offers set of widgets that can be added to the notebook. When serving notebook with Mercury, widget change triggers automatic re-execution of cells. Not all cells are re-executed, only cells with widget definition and below, so you can cache results from previous cells execution (loading large dataset or model).

Mercury comes with handy feautures to make sharing easy:

- decide to hide or show notebook's code,

- add authentication to notebooks, so only selected users can view them,

- export final notebook to PDF or HTML file,

- all to create output files in notebook, and make them downloadable,

- share multiple notebooks on one Site.

How does Mercury differ from existing solutions?

- it was desinged for notebooks, it offers simple re-execution of cells after widget update,

- it has built-in authentication.

Mercury is available on AGPLv3. We would like to offer a hosting service, to make deployment very easy (just upload notebook to have website). We offer commercial license for companies looking for private forks and dedicated support.

We’d love to hear from the HN community about this framework!