What problem does it solve? AI-generated code often lands as giant unreviewable commits with vague messages, mixed concerns, and no safety net when something breaks. This Skill imposes disciplined version control so every change is reviewable, revertible, and documented. ## Core Features & Use Cases - Atomic Commit Discipline: Enforces one logical change per commit (~100 lines), conventional message types (feat, fix, refactor, test, docs, chore), and separation of refactoring from feature work. - Branching & Worktrees: Guides trunk-based development with short-lived feature branches and git worktrees so multiple agents can work in parallel without interference. - Pre-Commit Hygiene & Debugging: Provides a pre-commit checklist (diff review, secret scanning, tests, lint, typecheck) plus git bisect, blame, and log techniques for debugging. - Use Case: An agent implementing a task-creation feature commits each slice separately (endpoint, form, wiring, tests), runs tests before each commit, and uses git reset --hard HEAD to recover instantly when a change breaks the build. ## Quick Start Ask the agent to apply the git workflow skill to commit the current changes as atomic, well-described commits after verifying tests pass.
