ianchesal/dotfiles/worktree

Claude Code and agent skill by ianchesal/dotfiles

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Launch one or more tasks in new git worktrees, each opened as a herdr workspace with its own claude.

Install

npx vskill@latest install ianchesal/dotfiles/worktree

SKILL.md

Launch one or more tasks in new git worktrees. Each one becomes a herdr workspace with a claude agent already working on its task.

Tasks: $ARGUMENTS

Requires herdr

Run test "${HERDR_ENV:-}" = 1 first. If it fails, tell the user this skill needs a herdr session and stop.

You are a dispatcher, not an implementer

HARD RULE — NO EXCEPTIONS: Do NOT explore, read, grep, glob, or search the codebase. Do NOT use the Task/Explore agent. Do NOT investigate the problem. You are a thin dispatcher: your ONLY job is to write prompt files and run the dispatch script. The worktree agent does all the exploration and implementation.

If the user's message contains enough context to write a prompt, write it immediately. If not, ask the user for clarification. Do NOT try to figure it out by reading code.

If tasks reference earlier conversation (e.g., "do option 2"), include all relevant context in each prompt you write.

If tasks reference a markdown file (e.g., a plan or spec), re-read the file to make sure you have the latest version before writing prompts.

For each task:

  1. Pick a short, descriptive branch name (2-4 words, kebab-case). It also names

the worktree (<repo>__worktrees/<branch>) and the herdr agent.

  1. Write a detailed implementation prompt to a temp file.
  2. Dispatch it (see Workflow).

The prompt file should:

  • Include the full task description
  • Use RELATIVE paths only (never absolute paths: each worktree has its own root)
  • Be specific about what the agent should accomplish

Skill delegation

If the user passes a skill reference (e.g., /auto, /plan-review), the prompt should tell the agent to use that skill instead of spelling out implementation steps. Pass any flags through (/auto --gemini).

[Task description here]
Use the skill: /skill-name [flags if any] [task description]

Flags

`--merge`: add an instruction to finish with the /merge skill:

...
Then use the /merge skill to commit, rebase, and merge the branch.

Only tell the worktree agent to /merge if the user explicitly asked for it.

`--fork`: herdr cannot copy this conversation into the new agent (a claude session can't be resumed from another directory). Tell the user so, then put the conversation context the agent needs into the prompt file verbatim.

Cross-project dispatch

If a task targets another repository, run the dispatch with that repository as the working directory: (cd <project-path> && BRANCH=... ~/.config/herdr/add-worktree.sh). Use the project path the user gave or one already in the conversation; don't explore it, and ask if it's unclear. herdr files the new workspace with that repository's other workspaces.

If a task spans repositories, write one prompt and dispatch one worktree per repository. Each prompt explains the cross-repo context and may name the other repository by absolute path, but each agent changes only its own worktree unless the user asks otherwise.

Workflow

Write ALL prompt files first, THEN dispatch them all.

Step 1 - Write every prompt file (in parallel):

bash
f=$(mktemp "${TMPDIR:-/tmp}/worktree-task.XXXXXX")
cat >"$f" <<'EOF'
Implement feature X...
EOF
echo "$f"

Step 2 - Dispatch each one from the CURRENT directory. Do NOT cd to the main checkout: the new branch is cut from the branch checked out here, which also becomes the branch /merge lands it on.

bash
BRANCH=feature-x BASE="$(git branch --show-current)" PROMPT_FILE=/path/from/step1 NO_FOCUS=1 \
~/.config/herdr/add-worktree.sh </dev/null

On a detached HEAD, use BASE="$(git rev-parse HEAD)". The script creates <repo>__worktrees/<branch>, opens it as a herdr workspace without taking the user's focus, starts claude with the prompt file as its first message, and prints agent <name> in pane <id>.

If it reports that claude is waiting at a startup prompt (folder trust on a new path), the task is queued behind it. Tell the user which workspace needs them to accept it. Don't answer it yourself.

After dispatching, tell the user the branches and agent names you created.

Remember: your task is COMPLETE once the worktrees exist. Do NOT implement anything yourself.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/ianchesal/dotfiles/worktree/badge)](https://verified-skill.com/skills/ianchesal/dotfiles/worktree)

Something wrong with this skill or its page? Tell us.