Jug 2.5.0 released — AI assistant skills, polars support, local config files

[Luis Pedro Coelho]

Hi all,

Jug 2.5.0 has been released. You can install/upgrade with:

pip install jug --upgrade

conda-forge version is also updated, so conda/mamba/pixi will all work
with the newer version.

Here are the highlights:

1. Project-local configuration files. Jug now discovers .jugrc or jugrc
files by walking up the directory tree from the current working
directory to the git project root. This lets you keep project-
specific Jug configuration alongside your code without polluting your
global config.

2. AI assistant integration. Jug now ships a bundled skill for AI
coding assistants (Claude Code, Codex). You can install it with a
single command:

# For Codex
jug install-skills --output ~/.codex/skills

# For Claude Code (user-level)
jug install-skills --output ~/.claude/skills

# For Claude Code (project-local)
jug install-skills --output .claude/skills

Once installed, you can ask the assistant to help you write jugfiles,
debug stuck tasks, or understand task dependencies. For example, in
Claude Code:

/jug write a jugfile for this pipeline /jug help me refactor this
workflow to use barriers safely

Or in Codex:

Use $jug to write a jugfile for this pipeline. Use $jug to debug why
these tasks are stuck waiting.

The skill includes CLI reference, common patterns, troubleshooting
guides, and example jugfiles. Use --force to overwrite a previously
installed copy.

3. Faster polars DataFrame saving. The file store now special-cases
polars DataFrames for significantly faster serialization.

4. More flexible boolean parsing. jug.options now accepts a wider range
of boolean representations.

Bugfixes:
- Fix _get_terminal_size_linux for Python 3.14 (patch by
justinrporter, GH #120)
- Fix dict_store backend for Python 3
- Fix describe in jug.task for Python 3

Full changelog: https://jug.readthedocs.io/en/latest/history.html

As always, bug reports and feedback are welcome at
https://github.com/luispedro/jug/issues I'd be particularly interested
in figuring out how the AI assistant skills could work better.


If you use Jug to generate results for a scientific publication,
please cite:

Coelho, L.P., (2017). Jug: Software for Parallel Reproducible
Computation in Python. Journal of Open Research Software.
5(1), p.30. https://doi.org/10.5334/jors.161



Best, Luis Pedro Coelho

---

Luis Pedro Coelho | Queensland University of Technology |
https://luispedro.org https://orcid.org/0000-0002-9280-7885