scientific-python-development

RSE Plugins

Custom AI agents and skills for Research Software Engineering (RSE) and Scientific Computing tasks, designed for use with Claude Code and compatible AI coding assistants.

GitHub Codespaces

Click the badge above to open a Codespace with all RSE plugins pre-installed. Claude Code and GitHub Copilot CLI are ready to use immediately.

Using a LiteLLM Gateway (Optional)

To route Copilot through a custom LiteLLM-compatible gateway, you need to set two secrets. Because GitHub Codespaces secrets can only be scoped to repos you own, follow these steps:

1. Fork this repo — Click Fork on github.com/uw-ssec/rse-plugins.
2. Add secrets — Go to your GitHub account settings: Settings → Codespaces → Secrets, then add:
   • LITELLM_BASE_URL — your gateway base URL
   • LITELLM_API_KEY — your gateway API key
   • When adding each secret, under Repository access, select your fork.
3. Open a Codespace from your fork — Go to your fork on GitHub and click Code → Codespaces → Create codespace on main.

The Codespace will automatically detect the secrets and configure Copilot to route through your gateway.

Purpose

This repository provides specialized agents and skills that understand the unique challenges of scientific software development, including:

• Modern Scientific Python development following community best practices
• Reproducible environment management with pixi
• Python packaging and distribution with pyproject.toml
• Comprehensive testing strategies with pytest
• Scientific computing workflows and numerical methods
• Research software engineering practices
• Domain-specific scientific computing (astronomy, geospatial analysis, climate science)
• Interactive data visualization with the HoloViz ecosystem (Panel, hvPlot, HoloViews, Datashader, GeoViews, Lumen)
• Scientific Python ecosystem (NumPy, Pandas, SciPy, Matplotlib, Xarray, Astropy, etc.)

Installation

To use these agents and skills in Claude Code, add this repository to your plugin marketplace:

/plugin marketplace add uw-ssec/rse-plugins

Once installed, the agents and skills will be available in your Claude Code environment and can be invoked when working on scientific software projects.

Available Plugins

The repository provides Claude Code plugins organized by domain. Each plugin contains agents (specialized AI personas) and skills (reusable knowledge modules).

Scientific Python Development Plugin

Expert agents and comprehensive skills for modern Scientific Python development.

Agents:

• Scientific Python Expert - Comprehensive agent for scientific Python development following Scientific Python Development Guide best practices
• Scientific Documentation Architect - Expert in creating comprehensive, user-friendly documentation for scientific software following Scientific Python community standards

Skills:

• pixi-package-manager - Fast, reproducible scientific Python environments with unified conda and PyPI management
• python-packaging - Modern packaging with pyproject.toml, src layout, and Hatchling build backend
• python-testing - Robust testing strategies with pytest following Scientific Python community guidelines
• code-quality-tools - Linting, formatting, and type checking tools for Python code quality
• scientific-documentation - Documentation best practices for scientific software including Sphinx, API docs, tutorials, and examples

When to use: Scientific computing projects, data analysis pipelines, research software development, package creation, reproducible research workflows

Scientific Domain Applications Plugin

Domain-specific scientific computing agents and skills for astronomy, geospatial analysis, climate science, and interactive visualization.

Agents:

• Astronomy & Astrophysics Expert - Expert in astronomical data analysis, FITS files, coordinate systems, and photometry/spectroscopy pipelines with Astropy

Skills:

• xarray-for-multidimensional-data - Work with labeled multidimensional arrays and NetCDF/Zarr datasets for climate and Earth science
• astropy-fundamentals - Astronomical data formats, coordinate transformations, physical units, and time handling with Astropy

When to use: Astronomy research, telescope data processing, climate data analysis, Earth science workflows, geospatial analysis

AI Research Workflows Plugin

Skills-first research workflows for Research Software Engineers and researchers — covering the full arc from understanding code and surveying prior art through planning, experimentation, implementation, validation, reproducibility, and handoff.