Contributing

Create a focused branch and pull request; do not push directly to main.
Explain the observable failure the rule fixes and include a before/after fixture in benchmark/cases.yml.
Rules must not infer AI authorship from style alone.
Do not add unsourced model-version claims or numeric thresholds.
Do not add code that executes reviewed content or downloads and runs remote code during review.
Run make check and make package before opening the pull request.
Keep skills/slop-cop/SKILL.md compact; put detailed examples and evidence in directly linked reference files.

Provide feedback