Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
64 lines (45 loc) · 2 KB

CONTRIBUTING.md

File metadata and controls

64 lines (45 loc) · 2 KB

Contributing

Thanks for your interest in improving the matplotlib skill.

Quick Start

git clone https://github.com/tvhahn/matplotlib-skill.git
cd matplotlib-skill
uv sync

Skill files live in skills/matplotlib/ (also available via symlink at .claude/skills/matplotlib/):

  • SKILL.md -- persona, workflow, code generation rules
  • style-reference.md -- full style spec (palettes, sizing, conventions)
  • patterns/P1-horizontal-bar.md through patterns/P9-pr-roc.md -- one file per chart type

Modifying a Pattern

  1. Edit the pattern file in skills/matplotlib/patterns/
  2. Run a clean-room test: scripts/chart-test-container.sh --quick
  3. Create a gallery entry in docs/gallery/NN-name/ following docs/gallery/TEMPLATE.md
  4. Include before/after PNGs in your PR

Adding a New Pattern

  1. Create patterns/P{N}-{name}.md following the structure of existing patterns
  2. Add it to the pattern index table in SKILL.md
  3. Add test prompts to scripts/chart-test-container.sh
  4. Run a full test: scripts/chart-test-container.sh

Style Invariants

These are immutable and must not change:

sns.set_theme(font_scale=1.0, style="whitegrid", font="DejaVu Sans")
sns.despine(left=True, bottom=True)

If you believe these should change, open an issue to discuss first.

Testing

# Docker clean room (primary verification)
scripts/chart-test-container.sh --quick            # 5 prompts, random from pools
scripts/chart-test-container.sh                    # all 9 prompts, random from pools
scripts/chart-test-container.sh --fixed            # all 9 prompts, original defaults

# Static analysis (fast, no Docker)
uv run scripts/evaluate_skill.py --check-renders

Pull Requests

  • One logical change per PR
  • Include before/after PNGs for any visual change
  • Run scripts/chart-test-container.sh --quick before submitting
  • Keep commit messages concise and descriptive

License

By contributing, you agree that your contributions will be licensed under the MIT License.