Guidelines for AI coding agents working on environments, the Docker images Deepnote runs notebooks in.
These images are what users get by default, so a change here is usually a user-visible change. Before opening a pull request, check whether it makes any documentation stale, and update it in the same pull request:
python/readme.md- the supported tags, what the images contain, and how to build on top of them. Keep the tag list accurate when versions are added or dropped.- The nearest readme next to what you changed (
gpu/,ir/,python/*) - User-facing documentation is not in this repository: it lives as Markdown under
docs/in the publicdeepnoterepository and is published at https://deepnote.com/docs. Changes to Python versions, pre-installed packages, or the default environment need a matching pull request there (docs/default-environment.md,docs/pre-installed-packages.md,docs/custom-environment.md).
Update only the documentation your change actually affects.
Keep each pull request as simple and clean as possible: one purpose per pull request, the smallest diff that achieves it, and no unrelated dependency bumps riding along with a behavior change.