worktrunk

Worktrunk

Use this skill when a user needs to configure wt, set up LLM commit messages, author project hooks, or run multiple parallel agent lanes with isolated worktrees.

Use this skill when

• User asks how to configure Worktrunk (wt config, ~/.config/worktrunk/config.toml, .config/wt.toml)
• User wants to set up LLM-generated commit messages or branch summaries
• User needs to author or debug hooks (post-start, pre-merge, etc.)
• User is setting up parallel agent workflows with wt switch --create --execute=…
• User asks about wt merge, wt step, or the merge pipeline
• User is debugging shell integration or worktree path layout

Do not use this skill when

• wt --version fails — Worktrunk is not installed; suggest brew install worktrunk
• A disposable clone is preferred over a worktree
• The task is primarily about creating a PR, watching checks, or post-push review flow — route to github-cli-pr-workflow

Routing boundary

Situation	Use this skill?	Route instead
wt installed; user wants hooks or config	Yes	—
User wants raw git worktree add/remove only	Yes	use this skill with raw git commands only
User needs PR/check workflow after pushing branch	No	github-cli-pr-workflow skill
wt --version fails	No	Suggest brew install worktrunk
Disposable clone, not a worktree	No	git clone directly
Implementation complete; deciding how to integrate branch	No	finishing-a-development-branch

Inputs to gather

Required before editing

• Is this config setup, hook authoring, or a parallel agent workflow?
• Does ~/.config/worktrunk/config.toml already exist? (wt config show)

Helpful if present

• Project type (Node, Rust, Python, etc.) for hook examples
• Whether LLM commit generation is desired and which LLM tool is available

First move

1. wt --version — confirm install (a current version is required; install via brew install worktrunk if missing)
2. wt config show — inspect active config and file locations
3. wt config create — scaffold user config if missing; wt config create --project for project config

Workflow

1. Configure worktree path — set worktree-path in user config (inside-repo .worktrees/ is recommended for agent harness worktree-manager compatibility)
2. Set up LLM commits — add [commit.generation] block; see assets/llm-commits-setup.md
3. Add project hooks — author .config/wt.toml hooks for install, dev server, DB, CI gates; see assets/hooks-reference.md
4. Enable branch summaries (optional) — [list] summary = true in user config
5. Run parallel agents (optional) — use wt switch --create --execute=<agent>; see assets/parallel-agents-recipes.md

Guardrails

• Must check wt --version before using any wt commands; fall back to raw git if Worktrunk is not installed.
• Must not bypass project hooks (post-start, pre-merge) without explicit justification.
• Should use wt merge for squash+rebase+cleanup in preference to manual git steps.
• Should configure [commit.generation] before relying on wt step commit for LLM messages.
• May use --no-squash when commit granularity matters for the PR.

Outputs

• A configured wt setup: user and/or project config files (~/.config/worktrunk/config.toml, .config/wt.toml), authored hooks, and optionally LLM commit generation enabled
• Working worktrees or parallel agent lanes as requested, verified with wt list and wt config show

Validation

wt config show                              # confirms settings loaded
wt switch --create test-wt-check           # creates worktree, fires hooks
wt list                                    # confirms branch with status markers
wt remove test-wt-check                   # cleans up

• Smoke test:
  • should trigger: "Configure wt hooks and parallel lanes for this repo."
  • should not trigger: "Create a raw git worktree for a one-off fix." (→ git-worktrees)

Examples

• Set up LLM commit messages for this repo:

  wt config show              # inspect current config location
  # add [commit.generation] block — see assets/llm-commits-setup.md for provider options
  wt switch --create test-llm-check
  # make a small change, then `wt merge` to confirm LLM message is generated
  

• Add a post-start hook that runs npm ci and starts the dev server:

  # .config/wt.toml
  [[hooks.post-start]]
  command = "npm ci && npm run dev -- --port {{.Port}}"
  

• Configure parallel agent lanes with unique ports per worktree:

  wt switch --create agent-lane-1 --execute="copilot agent start"
  wt switch --create agent-lane-2 --execute="copilot agent start"
  wt list   # shows both lanes with status markers
  

Reference files

• assets/hooks-reference.md — all hook types, template variables, filters, pipeline syntax
• assets/merge-pipeline.md — wt merge pipeline, flags, and wt step sub-commands
• assets/llm-commits-setup.md — LLM commit generation config for Claude Code, Codex, llm CLI, aichat
• assets/parallel-agents-recipes.md — one-shot alias pattern, dev server per worktree, DB per worktree, cold-start elimination
• references/config-reference.md — complete config key reference with defaults