git-workflow-and-versioning

To keep AI-generated code changes manageable and reversible, enforce a disciplined git workflow using trunk-based branching, atomic commits, and structured change summaries to track scope and intent across parallel development streams.

The best way to manage parallel AI coding branches is using git worktrees to isolate short-lived feature branches in separate directories, enabling multi-agent development without cross-branch interference while keeping the main branch deployable.

Write atomic commits for AI-assisted development by limiting each commit to one logical change, adding descriptive messages capturing intent and rationale, and running pre-commit verification like tests, linting, and type checks to ensure hygiene.

This git workflow approach requires a trunk-based branching strategy or adapted branching rules, utilizing short-lived feature branches to keep the main branch deployable while supporting release branches only when explicitly needed.

Structure change summaries for reviewing AI code modifications using a consistent format detailing what changed, what did not change, and remaining concerns to improve review quality and maintain strict scope discipline.