using-git-worktrees

name	using-git-worktrees
description	Use when starting feature work that needs isolation from current workspace or before executing implementation plans - creates isolated git worktrees with smart directory selection and safety verification

Using Git Worktrees

Overview

Git worktrees create isolated workspaces sharing the same repository, allowing work on multiple branches simultaneously without switching.

Core principle: Systematic directory selection + safety verification = reliable isolation.

Announce at start: "I'm using the using-git-worktrees skill to set up an isolated workspace."

Step 0: Discover Existing Convention

ALWAYS run this FIRST — before touching directories or names.

A repo may already have established worktree conventions created by teammates, other agent tools (Claude Desktop, CI), or prior sessions. Blindly applying defaults creates inconsistency and fights the existing pattern.

git worktree list

Examine the output. Look for:

1. Consistent parent directory — e.g., every existing worktree lives under .claude/worktrees/, .worktrees/, or similar. If ≥2 worktrees share a parent, that parent IS the convention. Use it.
2. Branch naming pattern — e.g., feature/<slug>, claude/<random>, wip/<topic>. If a pattern is visible in ≥2 existing branches, follow it.
3. Main repo location — the first entry (no branch decoration or tagged [main]) is the main worktree, not a candidate pattern.

If a convention is visible, use it and move directly to Step 2 (Safety Verification). Do NOT fall through to directory priority or ask the user — the pattern is already established.

If only one existing worktree exists and its convention is ambiguous (single data point), prefer CLAUDE.md over the sample-of-one.

If no existing worktrees (only the main repo is listed), fall through to Step 1.

Step 1: Directory Selection Process

Follow this priority order only when Step 0 found no convention:

1. Check Existing Directories

# Check in priority order
ls -d .claude/worktrees 2>/dev/null  # Claude-tool convention (if present)
ls -d .worktrees 2>/dev/null         # superpowers default (hidden)
ls -d worktrees 2>/dev/null          # Alternative

If found: Use that directory. Priority order is top-to-bottom; first match wins.

2. Check CLAUDE.md

grep -i "worktree.*director" CLAUDE.md 2>/dev/null

If preference specified: Use it without asking.

3. Ask User

If no directory exists and no CLAUDE.md preference:

No worktree directory found. Where should I create worktrees?

1. .worktrees/ (project-local, hidden)
2. ~/.config/superpowers/worktrees/<project-name>/ (global location)

Which would you prefer?

Step 2: Branch Naming Convention

Before creating the branch, determine its name. Priority:

1. Step 0 pattern — if existing worktrees show a branch prefix like feature/* or wip/*, use it.
2. CLAUDE.md rule — check CLAUDE.md for branch naming guidance:

   grep -iE "branch.*nam|feature/<slug>|worktree.*on branch" CLAUDE.md 2>/dev/null
   

   Many projects specify a pattern (e.g., feature/<slug> where slug derives from the plan filename).
3. Ask user if neither Step 0 nor CLAUDE.md gives a pattern.

Do NOT default to claude/<random> or similar agent-specific names in a repo that shows project-convention patterns — this creates inconsistency with teammate-created branches.

Safety Verification

For Project-Local Directories (.claude/worktrees, .worktrees, worktrees)

MUST verify directory is ignored before creating worktree:

# Check if directory is ignored (respects local, global, and system gitignore)
git check-ignore -q .claude/worktrees 2>/dev/null || \
  git check-ignore -q .worktrees 2>/dev/null || \
  git check-ignore -q worktrees 2>/dev/null

If NOT ignored:

Per Jesse's rule "Fix broken things immediately":

1. Add appropriate line to .gitignore
2. Commit the change
3. Proceed with worktree creation

Why critical: Prevents accidentally committing worktree contents to repository.

For Global Directory (~/.config/superpowers/worktrees)

No .gitignore verification needed - outside project entirely.

Creation Steps

1. Detect Project Name

project=$(basename "$(git rev-parse --show-toplevel)")

2. Create Worktree

# Determine full path — use the directory convention identified in
# Step 0 / Step 1, and the branch name from Step 2.
case $LOCATION in
  .claude/worktrees|.worktrees|worktrees)
    path="$LOCATION/$WORKTREE_NAME"
    ;;
  ~/.config/superpowers/worktrees/*)
    path="~/.config/superpowers/worktrees/$project/$WORKTREE_NAME"
    ;;
esac

# Note: WORKTREE_NAME (directory name under the parent) and BRANCH_NAME
# (full branch like "feature/<slug>") may differ. A common pattern is:
#   path: .claude/worktrees/rewrite
#   branch: feature/rewrite
# Match whatever Step 0 / Step 2 surfaced.

git worktree add "$path" -b "$BRANCH_NAME"
cd "$path"

3. Run Project Setup

Auto-detect and run appropriate setup:

# Node.js
if [ -f package.json ]; then npm install; fi

# Rust
if [ -f Cargo.toml ]; then cargo build; fi

# Python
if [ -f requirements.txt ]; then pip install -r requirements.txt; fi
if [ -f pyproject.toml ]; then
  if command -v uv >/dev/null 2>&1 && [ -f uv.lock ]; then
    uv sync
  else
    poetry install
  fi
fi

# Go
if [ -f go.mod ]; then go mod download; fi

4. Verify Clean Baseline

Run tests to ensure worktree starts clean:

# Examples - use project-appropriate command
npm test
cargo test
uv run pytest -q   # or: pytest
go test ./...

If tests fail: Report failures, ask whether to proceed or investigate.

If tests pass: Report ready.

5. Report Location

Worktree ready at <full-path>
Branch: <branch-name>
Tests passing (<N> tests, 0 failures)
Ready to implement <feature-name>

Quick Reference

Situation	Action
Existing worktrees show a convention	Follow it (Step 0)
.claude/worktrees/ exists	Use it (verify ignored)
.worktrees/ exists	Use it (verify ignored)
worktrees/ exists	Use it (verify ignored)
Multiple exist	First match in priority order
None exist	Check CLAUDE.md → Ask user
Directory not ignored	Add to .gitignore + commit
Tests fail during baseline	Report failures + ask
No package.json/Cargo.toml/pyproject.toml	Skip dependency install

Common Mistakes

Skipping convention discovery

• Problem: Creates worktrees that conflict with teammate or other-tool conventions; branches show up with inconsistent naming in git branch
• Fix: Always run git worktree list BEFORE directory selection; follow established patterns

Skipping ignore verification

• Problem: Worktree contents get tracked, pollute git status
• Fix: Always use git check-ignore before creating project-local worktree

Assuming directory location

• Problem: Creates inconsistency, violates project conventions
• Fix: Follow priority: discovered > existing > CLAUDE.md > ask

Hardcoding branch-name patterns

• Problem: Tool-specific defaults (e.g., claude/<random>) clash with teammate-set patterns like feature/<slug>
• Fix: Infer from existing branches + CLAUDE.md; only fall back to defaults when nothing is set

Proceeding with failing tests

• Problem: Can't distinguish new bugs from pre-existing issues
• Fix: Report failures, get explicit permission to proceed

Hardcoding setup commands

• Problem: Breaks on projects using different tools
• Fix: Auto-detect from project files (package.json, etc.)

Example Workflow

You: I'm using the using-git-worktrees skill to set up an isolated workspace.

[Step 0: git worktree list shows 3 existing worktrees under
 .claude/worktrees/ with branches feature/*]
[Convention identified: .claude/worktrees/<slug> + feature/<slug>]
[Verify .claude/worktrees/ is in .gitignore - confirmed]
[Plan slug: "rewrite" → branch: feature/rewrite, path: .claude/worktrees/rewrite]
[Create worktree: git worktree add .claude/worktrees/rewrite -b feature/rewrite]
[uv sync]
[uv run pytest -q — 720 passing]

Worktree ready at /Users/jesse/myproject/.claude/worktrees/rewrite
Branch: feature/rewrite
Tests passing (720 tests, 0 failures)
Ready to implement rewrite feature

Red Flags

Never:

• Skip git worktree list before selecting a directory
• Create worktree without verifying it's ignored (project-local)
• Skip baseline test verification
• Proceed with failing tests without asking
• Assume directory location when ambiguous
• Skip CLAUDE.md check
• Use agent-default branch naming (claude/<random>) when a project convention is visible

Always:

• Discover existing convention first (Step 0)
• Follow directory priority: discovered > existing > CLAUDE.md > ask
• Verify directory is ignored for project-local
• Match branch naming to convention (Step 0 → CLAUDE.md → ask)
• Auto-detect and run project setup
• Verify clean test baseline

Integration

Called by:

• brainstorming (Phase 4) - REQUIRED when design is approved and implementation follows
• subagent-driven-development - REQUIRED before executing any tasks
• executing-plans - REQUIRED before executing any tasks
• Any skill needing isolated workspace

Pairs with:

• finishing-a-development-branch - REQUIRED for cleanup after work complete