My git workflow with Claude Code (keeping history clean)

#claudecode #git #productivity #webdev

When Claude Code is doing most of the implementation, git history gets messy fast. You end up with commits like "fix the thing", "actually fix the thing", "revert fix and try different approach".

Here's the workflow I settled on.

Work on branches, always

Never let Claude commit directly to main. This is more important with AI-assisted development because Claude will confidently make changes that seem reasonable but need review.

git checkout -b feature/[what-we're-building]

Tell Claude the commit message style before starting

Add to your CLAUDE.md:

When committing: use conventional commits format.
feat: for new features
fix: for bug fixes
refactor: for refactoring
test: for test-only changes

Keep subjects under 72 chars. No period at end.

Without this, Claude writes commit messages that are paragraphs.

Atomic commits

Claude's default is to batch everything into one commit. Prompt: "Commit each logical change separately. If you're changing the data model and updating the UI, those are two commits."

Review before push

Always run git diff main before pushing, even when you watched Claude make the changes. This catches:

Accidental deletions Claude didn't mention
Debug code that wasn't cleaned up
Changes to files that were out of scope
Hardcoded values that should be config

PR descriptions

Write a PR description for this branch.
Include:
- What changed (bullet points, not prose)
- Why it changed
- How to test it

Audience: the person reviewing this who didn't see our conversation.

The last line is key. Without it you get PR descriptions that reference "what we discussed" — meaningless to a reviewer looking at it cold.

The CLAUDE.md git config

## Git rules
- Never commit to main directly
- Always use conventional commits
- Atomic commits — one logical change per commit
- Run tests before committing
- No "WIP" commits on branches that will be PRed