setup-sprint

# Setup Sprint Prepare parallel Git worktrees for a batch of issues filtered by label. This skill sets up the workspace but does not start Claude Code instances or begin implementation. ## Your task ### Step 1: Verify starting point - Confirm the current branch is `main` (or the repository's default branch). If not, ask the user before proceeding. - Run `git pull` to ensure `main` is up to date. Worktrees branch from the current HEAD, so starting from stale history creates merge headaches later. ### Step 2: Find issues - Determine the project name from the Git repository root directory name: ```bash project_name="$(basename "$(git rev-parse --show-toplevel)")" ``` This ensures the correct name even when running from a subdirectory. - Run `gh issue list --label $ARGUMENTS --state open --json number,title,labels` to find all open issues with the specified label. - Present the list to the user. **CHECKPOINT**: Confirm which issues to include and in what order. The user may exclude issues (e.g., if two touch the same files and should be done sequentially) or adjust the batch. ### Step 3: Create worktrees For each confirmed issue: - Determine the branch prefix from the issue's type label: - `feat/` for issues labeled `feat` - `fix/` for issues labeled `fix` - `chore/` for issues labeled `chore` - `docs/` for issues labeled `docs` - `test/` for issues labeled `test` - If no type label is present, infer from context (default to `chore/`) - Generate a ...