git-workflow-and-versioning

What problem does it solve? AI agents and developers generating code at high speed often produce giant unreviewable commits, vague messages, and diverging long-lived branches. This Skill enforces disciplined version control so every change stays reviewable, revertable, and traceable. ## Core Features & Use Cases - Atomic Commit Discipline: Enforces small, single-purpose commits (~100 lines) with conventional message types (feat, fix, refactor, test, docs, chore) that explain the why. - Trunk-Based Branching Strategy: Guides short-lived feature branches merged within 1-3 days, with naming conventions and feature-flag alternatives to long-lived branches. - Parallel Work with Worktrees: Shows how to use git worktrees so multiple agents work on separate branches simultaneously without interference. - Use Case: An AI agent implements a feature across several slices, committing each verified increment with a descriptive message, so a broken change can be reverted instantly via the save point pattern. ## Quick Start Ask the agent to commit the current changes following atomic commit conventions with a descriptive conventional-commit message.