BestieTemplate.jl
Your best practices friend.
What does BestieTemplate do?
Creating Julia packages involve the creation and edition of many tiny files.
Wouldn't it be great to automate this?
This is exactly what BestieTemplate does.
FAQ
- How is
BestieTemplatedifferent fromPkgTemplates?- it can be applied to existing packages
- it invites to follow some (opinionated) best practices
- it can be reapplied to acquire updates made to the template
- it is automatically reapplied through Pull Requests made by the
Copier.ymlworkflow (Work in progress)
Architecture
Under the hood, BestieTemplate is no more and no less than:
- a copier template/skeleton for Julia packages (see folder template); and
- a package that wraps
copierin Julia usingPythonCallwith some convenience functions.
Quickstart
Install BestieTemplate in your chosen environment (we recommend globally) by entering pkg mode by pressing ] and then:
julia> # press ]
pkg> add BestieTemplate
then:
julia> using BestieTemplate
julia> BestieTemplate.generate("path/to/YourNewPackage.jl")
julia> # or BestieTemplate.apply("path/to/YourExistingPackage.jl")
please note that "YourPackage.jl" can either be a fresh new package or an existing one.
If you like what you see, check the full usage guide.
Without Julia, or from an AI coding agent
Individual features of the template (AGENTS.md, changelog, Dependabot, pre-commit, …) can be added to an existing package without installing Julia, using the experimental bestie-template Python package.
With uv installed, no install step is needed:
uvx --from bestie-template bestie list-features
uvx --from bestie-template bestie add-feature changelog,dependabot path/to/MyPackage.jl
See the quick guide for more details, and its add_feature counterpart for the Julia version.
For AI coding agents, the same workflow is packaged as the bestie-features skill, which is installable with
npx skills add JuliaBesties/BestieTemplate.jl
See skills.sh for what agent skills are and which agents support them.
Users and Examples
The following are users and examples of repos using this template, or other templates based on it. Feel free to create a pull request to add your repo.
- This package itself uses the template.
- TulipaIO.jl
Contributing
If you would like to get involved in the BestieTemplate growth, please check our contributing guide. We welcome contributions of many types, including coding, reviewing, creating issues, creating tutorials, interacting with users, etc. Make sure to follow our code of conduct.
If your interest is in developing the package, check the development guide as well.
AI Coding Assistant Attribution
We use and accepts pull requests with AI coding assistants to help with development, but we expect the committers to understand and be responsible for the code that they introduce. All commits that receive AI assistance should be signed off with:
Co-authored-by: MODEL NAME (FULL MODEL VERSION) <EMAIL>
For example:
Co-authored-by: Claude Code (claude-sonnet-4-20250514) <[email protected]>
References
Here is a list of links/repos that include content that we have used for inspiration, or used directly. This is most likely not a complete list, since many of the things included here were based on existing packages and knowledge that we brought from other projects. This also doesn't explain where each file came from or why they are here. You can find some of that information in the Explanation section of the docs.
- PkgTemplates.jl, naturally. We used it for many years, and in particular for the initial TulipaEnergyModel.jl commit (see below).
- Netherlands eScience Center's python template includes many of the best practices that we apply here. We used many of the ideas there in a Julia context, and took many non-Julia specific ideas from there.
- TulipaEnergyModel.jl was the project that motivated this version of a template. From the start we decide to implement many best practices and so we started from a PkgTemplates.jl template and started adding parts of the python template that made sense.
- The Julia Smooth Optimizers package ecosystem was one of the main motivations to look for a solution that could be applied and reapplied to existing packages, in particular to help maintainers.
