ai-conventional-commit-cli✂️ Splitting Commits
Cluster a large staged diff into multiple atomic commits, each selectively staged and executed.
When a staged diff spans several unrelated changes, split clusters its hunks
semantically and proposes a series of atomic commits. Each candidate is staged
selectively and executed in turn.
⚡ Basic usage
ai-conventional-commit split
# 1. refactor(parser): simplify token scanning
# 2. feat(parser): support negated glob segments
# 3. test(parser): add cases for brace + extglob combos🎯 Target a specific number of commits
Pass n to ask the splitter for that many commits:
ai-conventional-commit split 2⚙️ How it works
The splitter reads the staged diff, groups related hunks into clusters, and generates one Conventional Commit candidate per cluster. On confirmation, each candidate has its hunks staged selectively before the commit is executed — so the history ends up as a clean series of focused, atomic commits rather than one large mixed change.
Style, model, and privacy settings apply to split exactly as they do to a
single generate. See Styles & Gitmoji
and Privacy & Tiers.