scaryrawr/worktrunk

Worktrunk Skill

Use this skill when users need to parallelize work into independent chunks using git worktrees managed by wt.

Core commands

• wt switch --create <chunk-branch> (or wt switch -c <chunk-branch>) to create and switch to a new worktree
• wt switch <chunk-branch> to switch to an existing worktree
• wt list to inspect all managed worktrees
• wt remove to clean up the current worktree and branch when finished

Mandatory preflight before creation

Before any wt switch --create ... command, run:

scripts/check-disk-fit.sh -p .

Rules:

• If fits=false or the script exits non-zero, do not create the worktree.
• Report estimated_total_bytes, free_bytes, safety_margin_bytes, and required_bytes to the user.
• Only proceed with wt switch --create ... when the check exits 0.

For estimate-only requests (without creation), run:

scripts/estimate-worktree-footprint.sh -p .

Chunking guidance for parallel agents

• Split requests into narrowly scoped chunks that can be reviewed and merged independently.
• Use one branch/worktree per chunk, with clear names (e.g., feat-auth-token-refresh).
• Avoid overlapping edits across active chunks when possible to reduce merge conflicts.
• Keep each chunk shippable: code + tests/docs relevant to that chunk.

Parallel execution pattern

1. Define chunk list and naming.
2. For each chunk: run disk preflight check, then create worktree.
3. Launch agent work in each chunked worktree.
4. Review and commit each chunk independently.
5. Merge and clean up completed chunks with wt remove.

Prerequisite checks

Before using wt, verify:

command -v wt >/dev/null 2>&1
wt --help >/dev/null

If wt is missing, instruct the user to install Worktrunk first.