What problem does it solve?
Committing a large or continuously changing Git worktree is risky: unrelated changes get mixed, generated artifacts sneak in, concurrent edits get clobbered, and the shared index cannot be trusted. This Skill turns that chaos into a disciplined campaign of reviewable, well-described commits without ever destroying work.
Core Features & Use Cases
- Rolling commit campaigns: Reconnoiter, classify into green/yellow/red confidence lanes, and drain coherent change packets through a serial commit authority until every path is committed or has an evidence-backed exclusion.
- Private-index transactions: Stage each commit in a temporary index and advance HEAD with compare-and-swap via
git commit-tree and git update-ref, so concurrent work in the shared checkout is never overwritten.
- Artifact rejection and residual audit: Exclude build output, caches, QA captures, secrets, and oversized binaries with evidence, then close out only after every residual path is classified.
- Use Case: You return to a checkout with hundreds of uncommitted files from days of parallel work. The Skill groups them by intent, writes Conventional Commit messages, validates each packet, and commits them in dependency order while leaving genuinely uncertain material uncommitted with reasons.
Quick Start
Ask the AI to analyze this dirty worktree and commit all of my pending changes as coherent, reviewable commits using the happier-commit-worktree skill.
