What problem does it solve?
When feature work or plan execution must not affect your main working directory, you need reliable isolation that prevents accidental commits and keeps setup reproducible.
Core Features & Use Cases
- Directory Selection Priority: Chooses a worktree base directory using a clear order (existing hidden or visible worktrees directory first, then AGENTS.md, then legacy CLAUDE.md, otherwise asks the user).
- Safety Verification for Project-Local Worktrees: Ensures project-local worktree directories are ignored via git check-ignore before creating them, and if not ignored it instructs you to update .gitignore, commit, and only then proceed.
- Automated Bootstrap + Baseline Testing: Creates a new git worktree on an appropriate branch, runs project-appropriate dependency installation, and verifies a clean baseline by running the most relevant test command(s) detected by file presence.
Use case example: Before implementing an approved plan, you can create an isolated worktree tied to a dedicated branch, automatically install the needed dependencies for a Node/Rust/Python/Go project, and confirm tests pass before starting the real work.
Quick Start
Tell your AI agent to use the using-git-worktrees skill to create an isolated Git worktree, run the project setup, and verify the baseline tests before beginning implementation.