Worktree Parallel

Workflow

1. Confirm the repo root (use current working directory unless the user specifies another repo).
2. Determine a feature slug and branch name.
   • Default branch pattern: feature/<slug>.
   • If the user already has a branch name, use it.
3. Choose a base directory for linked worktrees.
   • Default for this repo: ../<project_name>-wt (create it if missing).
4. Create the worktree.
   • New branch: git worktree add -b feature/<slug> <base>/<slug>
   • Existing branch: git worktree add <base>/<slug> <branch>
5. Bootstrap the new worktree dependencies.
   • cd <path> && pnpm run setup-worktree
   • If setup fails, resolve bootstrap issues before opening the parallel session.
6. Open the worktree in the editor.
   • Use code -n <path> (Cursor supports the code CLI).
   • If code is unavailable, ask for the preferred editor command (e.g., cursor -n, code, or another).
7. Suggest a parallel Codex session in the new worktree.
   • cd <path> && codex
8. Confirm with git worktree list if needed.
9. If the new session will run tests, start and use a dev server from that same worktree only, then stop it after tests.

Safety

• Do not use --force unless the user explicitly asks.
• Do not remove existing worktrees unless the user requests cleanup.
• Do not stop dev servers you did not start.

Optional Cleanup (only if asked)

• git worktree remove <path>
• git branch -d <branch>