Claude-Skills git-worktree-manager
git clone https://github.com/borghei/Claude-Skills
T=$(mktemp -d) && git clone --depth=1 https://github.com/borghei/Claude-Skills "$T" && mkdir -p ~/.claude/skills && cp -r "$T/engineering/git-worktree-manager" ~/.claude/skills/borghei-claude-skills-git-worktree-manager && rm -rf "$T"
engineering/git-worktree-manager/SKILL.mdGit Worktree Manager
Tier: POWERFUL Category: Engineering / Developer Tooling Maintainer: Claude Skills Team
Overview
Manage parallel development workflows using Git worktrees with deterministic naming, automatic port allocation, environment file synchronization, dependency installation, and cleanup automation. Optimized for multi-agent workflows where each agent or terminal session owns an isolated worktree with its own ports, environment, and running services.
Keywords
git worktree, parallel development, branch isolation, port allocation, multi-agent development, worktree cleanup, Docker Compose worktree, concurrent branches
Core Capabilities
1. Worktree Lifecycle Management
- Create worktrees from new or existing branches with deterministic naming
- Copy .env files from main repo to new worktrees
- Install dependencies based on lockfile detection
- List all worktrees with status (clean/dirty, ahead/behind)
- Safe cleanup with uncommitted change detection
2. Port Allocation
- Deterministic port assignment per worktree (base + index * stride)
- Collision detection against running processes
- Persistent port map in
.worktree-ports.json - Docker Compose override generation for per-worktree ports
3. Multi-Agent Isolation
- One branch per worktree, one agent per worktree
- No shared state between agent workspaces
- Conflict-free parallel execution
- Task ID mapping for traceability
4. Cleanup Automation
- Stale worktree detection by age
- Merged branch detection for safe removal
- Dirty state warnings before deletion
- Bulk cleanup with safety confirmations
When to Use
- You need 2+ concurrent branches open with running dev servers
- You want isolated environments for feature work, hotfixes, and PR review
- Multiple AI agents need separate workspaces that do not interfere
- Your current branch is blocked but a hotfix is urgent
- You want automated cleanup instead of manual
operationsrm -rf
Quick Start
Create a Worktree
# Create worktree for a new feature branch git worktree add ../wt-auth -b feature/new-auth main # Create worktree from an existing branch git worktree add ../wt-hotfix hotfix/fix-login # Create worktree in a dedicated directory git worktree add ~/worktrees/myapp-auth -b feature/auth origin/main
List All Worktrees
git worktree list # Output: # /Users/dev/myapp abc1234 [main] # /Users/dev/wt-auth def5678 [feature/new-auth] # /Users/dev/wt-hotfix ghi9012 [hotfix/fix-login]
Remove a Worktree
# Safe removal (fails if there are uncommitted changes) git worktree remove ../wt-auth # Force removal (discards uncommitted changes) git worktree remove --force ../wt-auth # Prune stale metadata git worktree prune
Port Allocation Strategy
Deterministic Port Assignment
Each worktree gets a block of ports based on its index:
Worktree Index App Port DB Port Redis Port API Port ──────────────────────────────────────────────────────────────── 0 (main) 3000 5432 6379 8000 1 (wt-auth) 3010 5442 6389 8010 2 (wt-hotfix) 3020 5452 6399 8020 3 (wt-feature) 3030 5462 6409 8030
Formula:
port = base_port + (worktree_index * stride)
Default stride: 10
Port Map File
Store the allocation in
.worktree-ports.json at the worktree root:
{ "worktree": "wt-auth", "branch": "feature/new-auth", "index": 1, "ports": { "app": 3010, "database": 5442, "redis": 6389, "api": 8010 }, "created": "2026-03-09T10:30:00Z" }
Port Collision Detection
# Check if a port is already in use check_port() { local port=$1 if lsof -i :"$port" > /dev/null 2>&1; then echo "PORT $port is BUSY" return 1 else echo "PORT $port is FREE" return 0 fi } # Check all ports for a worktree for port in 3010 5442 6389 8010; do check_port $port done
Full Worktree Setup Script
#!/bin/bash # setup-worktree.sh — Create a fully prepared worktree set -euo pipefail BRANCH="${1:?Usage: setup-worktree.sh <branch-name> [base-branch]}" BASE="${2:-main}" WT_NAME="wt-$(echo "$BRANCH" | sed 's|.*/||' | tr '[:upper:]' '[:lower:]')" WT_PATH="../$WT_NAME" MAIN_REPO="$(git rev-parse --show-toplevel)" echo "Creating worktree: $WT_PATH from $BASE..." # 1. Create worktree if git rev-parse --verify "$BRANCH" > /dev/null 2>&1; then git worktree add "$WT_PATH" "$BRANCH" else git worktree add "$WT_PATH" -b "$BRANCH" "$BASE" fi # 2. Copy environment files for envfile in .env .env.local .env.development; do if [ -f "$MAIN_REPO/$envfile" ]; then cp "$MAIN_REPO/$envfile" "$WT_PATH/$envfile" echo "Copied $envfile" fi done # 3. Allocate ports WT_INDEX=$(git worktree list | grep -n "$WT_PATH" | cut -d: -f1) WT_INDEX=$((WT_INDEX - 1)) STRIDE=10 cat > "$WT_PATH/.worktree-ports.json" << EOF { "worktree": "$WT_NAME", "branch": "$BRANCH", "index": $WT_INDEX, "ports": { "app": $((3000 + WT_INDEX * STRIDE)), "database": $((5432 + WT_INDEX * STRIDE)), "redis": $((6379 + WT_INDEX * STRIDE)), "api": $((8000 + WT_INDEX * STRIDE)) } } EOF echo "Ports allocated (index $WT_INDEX)" # 4. Update .env with allocated ports if [ -f "$WT_PATH/.env" ]; then APP_PORT=$((3000 + WT_INDEX * STRIDE)) DB_PORT=$((5432 + WT_INDEX * STRIDE)) sed -i.bak "s/APP_PORT=.*/APP_PORT=$APP_PORT/" "$WT_PATH/.env" sed -i.bak "s/:5432/:$DB_PORT/g" "$WT_PATH/.env" rm -f "$WT_PATH/.env.bak" echo "Updated .env with worktree ports" fi # 5. Install dependencies cd "$WT_PATH" if [ -f "pnpm-lock.yaml" ]; then pnpm install --frozen-lockfile elif [ -f "package-lock.json" ]; then npm ci elif [ -f "yarn.lock" ]; then yarn install --frozen-lockfile elif [ -f "requirements.txt" ]; then pip install -r requirements.txt elif [ -f "go.mod" ]; then go mod download fi echo "" echo "Worktree ready: $WT_PATH" echo "Branch: $BRANCH" echo "App port: $((3000 + WT_INDEX * STRIDE))" echo "" echo "Next: cd $WT_PATH && pnpm dev"
Docker Compose Per-Worktree
# docker-compose.worktree.yml — override for worktree-specific ports # Usage: docker compose -f docker-compose.yml -f docker-compose.worktree.yml up services: postgres: ports: - "${DB_PORT:-5432}:5432" environment: POSTGRES_DB: "myapp_${WT_NAME:-main}" redis: ports: - "${REDIS_PORT:-6379}:6379" app: ports: - "${APP_PORT:-3000}:3000" environment: DATABASE_URL: "postgresql://dev:dev@postgres:5432/myapp_${WT_NAME:-main}"
Launch with worktree-specific ports:
DB_PORT=5442 REDIS_PORT=6389 APP_PORT=3010 WT_NAME=auth \ docker compose -f docker-compose.yml -f docker-compose.worktree.yml up -d
Cleanup Automation
#!/bin/bash # cleanup-worktrees.sh — Safe worktree cleanup set -euo pipefail STALE_DAYS="${1:-14}" DRY_RUN="${2:-true}" echo "Scanning worktrees (stale threshold: ${STALE_DAYS} days)..." echo "" git worktree list --porcelain | while read -r line; do case "$line" in worktree\ *) WT_PATH="${line#worktree }" ;; branch\ *) BRANCH="${line#branch refs/heads/}" # Skip main worktree if [ "$WT_PATH" = "$(git rev-parse --show-toplevel)" ]; then continue fi # Check if branch is merged MERGED="" if git branch --merged main | grep -q "$BRANCH" 2>/dev/null; then MERGED=" [MERGED]" fi # Check for uncommitted changes DIRTY="" if [ -d "$WT_PATH" ]; then cd "$WT_PATH" if [ -n "$(git status --porcelain)" ]; then DIRTY=" [DIRTY - has uncommitted changes]" fi cd - > /dev/null fi # Check age if [ -d "$WT_PATH" ]; then AGE_DAYS=$(( ($(date +%s) - $(stat -f %m "$WT_PATH" 2>/dev/null || stat -c %Y "$WT_PATH" 2>/dev/null)) / 86400 )) STALE="" if [ "$AGE_DAYS" -gt "$STALE_DAYS" ]; then STALE=" [STALE: ${AGE_DAYS} days old]" fi fi echo "$WT_PATH ($BRANCH)$MERGED$DIRTY$STALE" if [ -n "$MERGED" ] && [ -z "$DIRTY" ] && [ "$DRY_RUN" = "false" ]; then echo " -> Removing merged clean worktree..." git worktree remove "$WT_PATH" fi ;; esac done echo "" git worktree prune echo "Done. Run with 'false' as second arg to actually remove."
Multi-Agent Workflow Pattern
When running multiple AI agents (Claude Code, Cursor, Copilot) on the same repo:
Agent Assignment: ─────────────────────────────────────────────────── Agent 1 (Claude Code) → wt-feature-auth (port 3010) Agent 2 (Cursor) → wt-feature-billing (port 3020) Agent 3 (Copilot) → wt-bugfix-login (port 3030) Main repo → integration (main) (port 3000) ─────────────────────────────────────────────────── Rules: - Each agent works ONLY in its assigned worktree - No agent modifies another agent's worktree - Integration happens via PRs to main, not direct merges - Port conflicts are impossible due to deterministic allocation
Decision Matrix
| Scenario | Action |
|---|---|
| Need isolated dev server for a feature | Create a new worktree |
| Quick diff review of a branch | in current tree (no worktree needed) |
| Hotfix while feature branch is dirty | Create dedicated hotfix worktree |
| Bug triage with reproduction branch | Temporary worktree, cleanup same day |
| PR review with running code | Worktree at PR branch, run tests |
| Multiple agents on same repo | One worktree per agent |
Validation Checklist
After creating a worktree, verify:
shows the expected path and branchgit worktree list
exists with unique port assignments.worktree-ports.json
files are present and contain worktree-specific ports.env
(or equivalent) completed without errorspnpm install- Dev server starts on the allocated port
- Database connects on the allocated DB port
- No port conflicts with other worktrees or services
Common Pitfalls
- Creating worktrees inside the main repo directory — always use
to keep them alongside../wt-name - Reusing localhost:3000 across all branches — causes port conflicts; use deterministic allocation
- Sharing one DATABASE_URL across worktrees — each needs its own database or schema
- Removing a worktree with uncommitted changes — always check dirty state before removal
- Forgetting to prune after branch deletion — run
to clean metadatagit worktree prune - Not updating .env ports after worktree creation — the setup script should handle this automatically
Best Practices
- One branch per worktree, one agent per worktree — never share
- Keep worktrees short-lived — remove after the branch is merged
- Deterministic naming — use
pattern for easy identificationwt-<topic> - Persist port mappings — store in
, not in memory.worktree-ports.json - Run cleanup weekly — scan for stale and merged-branch worktrees
- Include worktree path in terminal title — prevents wrong-window commits
- Never force-remove dirty worktrees — unless changes are intentionally discarded
Troubleshooting
| Problem | Cause | Solution |
|---|---|---|
| Branch is already active in another worktree | Use to find where the branch is checked out, then switch to a different branch or remove the existing worktree first |
| Port conflict despite deterministic allocation | A non-worktree process is occupying the assigned port | Run to identify the process, terminate it or adjust the stride/base in the port allocation formula |
file missing after worktree creation | Setup script was not run or does not exist in the main repo | Copy manually from the main repo root, or re-run which handles env file copying |
reports nothing but stale paths remain | Worktree directory was deleted manually without | Run to clean orphaned metadata, then verify with |
| Dependencies fail to install in new worktree | Lockfile references a private registry or cache not available in the worktree path | Ensure , , or pip config files are copied alongside during setup |
| Docker Compose services start on wrong ports | The override was not included in the compose command | Always pass both files: |
| Worktree shows as dirty immediately after creation | Untracked files from copy or generated | Add and copied env files to in the project |
Success Criteria
- Zero port conflicts across all active worktrees measured by
checks returning no collisions after setuplsof - Worktree creation under 60 seconds including dependency installation for projects with warm package caches
- 100% env parity between main repo and worktrees verified by diffing
keys (values may differ for ports).env - Stale worktree count stays at zero when cleanup automation runs on a weekly schedule with a 14-day threshold
- No cross-worktree interference validated by running concurrent dev servers in 3+ worktrees simultaneously without failures
- Branch-to-worktree traceability maintained via
present in every active worktree with correct metadata.worktree-ports.json - Cleanup safety rate of 100% meaning no worktree with uncommitted changes is ever removed without explicit
confirmation--force
Scope & Limitations
This skill covers:
- Git worktree lifecycle: creation, listing, status inspection, and removal
- Deterministic port allocation and collision avoidance for parallel dev servers
- Environment file synchronization and Docker Compose override patterns
- Multi-agent workspace isolation strategies and cleanup automation
This skill does NOT cover:
- Git branching strategies or merge conflict resolution (see
andpr-review-expert
)release-manager - Secret rotation, vault integration, or credential management (see
)env-secrets-manager - CI/CD pipeline configuration or automated test orchestration (see
)ci-cd-pipeline-builder - Monorepo package management, workspace linking, or cross-package dependency resolution (see
)monorepo-navigator
Integration Points
| Skill | Integration | Data Flow |
|---|---|---|
| Worktree setup copies files that contain secrets managed by this skill | files flow from main repo to each worktree; secret references remain consistent across all copies |
| CI pipelines can spin up worktrees for parallel test matrix execution | Pipeline config triggers per matrix job; port allocation prevents service collisions |
| Release branches get dedicated worktrees for stabilization while feature work continues | Release worktree is created from the release branch; merged status drives cleanup automation |
| In monorepo setups, worktrees must respect package boundaries and shared dependencies | Worktree creation inherits the monorepo root lockfile; package-level dev servers use allocated port blocks |
| PR reviews can be performed in isolated worktrees with running code for manual validation | Reviewer creates a worktree at the PR branch, runs the dev server on allocated ports, and removes after review |
| Stale worktrees and abandoned branches surface as tech debt indicators | Cleanup script output feeds into debt tracking; worktree age and merge status inform priority scores |