Create small atomic git commits using conventional commit format at validated checkpoints.
npx skills add https://github.com/jeremysball/dotfiles --skill commit-jeremysball
Please help me install this Agent Skill. Skill: commit Source: https://github.com/jeremysball/dotfiles/tree/main/.pi/skills-archive/commit Command: npx skills add https://github.com/jeremysball/dotfiles --skill commit-jeremysball
git add -p for mixed hunks. - Conventional Commit Format: Standardizes messages as type[scope]: description with types like feat, fix, docs, refactor, and test. - Use Case: After finishing a feature with tests and docs updated, stage only the related files, run targeted validation, and commit as feat(memory): add support profile scope contract without dragging in unrelated changes. ## Quick Start Ask the assistant to review the current worktree and commit the completed changes as small conventional commits.High-intent search queries and answers about installing and using this skill.
Use the format type[scope]: description, such as feat(auth): add login endpoint. Keep the type and scope lowercase, write an imperative description, and keep the summary under 72 characters.
Group modified files by logical change rather than directory, name each commit in one line, then stage and commit each slice separately. Use git add -p when a single file contains unrelated hunks.
Use git add -p only when one file contains unrelated hunks that belong to different commits. If the change is already clean and atomic, plain git add of the intended files is sufficient.
Confirm the staged diff is one logical change, relevant validation passed on the touched surface, no unrelated files are staged, and the message is specific. Review git diff --cached before committing.
First determine whether everything is one logical change or several. If it is several slices, propose the split and get confirmation; only make one large commit if the user explicitly insists.
thangchungImplements multi-file code changes in tested, committed vertical slices.
sumrae412Automate conventional commits, pull requests, and 10-step code review before merging.
DataDogRun Hatch-based lint, format, and quality checks on modified files.
paywherebSquash all branch commits into one via interactive git rebase.