Worker Worktree Skill

"Parallel development without conflicts – each agent works in their own worktree."

Verification Checklist

Before starting work:

• Worktree exists (git worktree list shows your worktree)
• In worktree directory (pwd shows ../{agent}-worktree)
• On correct branch (git branch shows {agent}-worktree)
• Main branch merged (git merge origin/main completed)
• No merge conflicts (git status is clean)
• Latest changes pulled (git log shows recent main commits)

Quick Reference Commands

Action	Command
List worktrees	`git worktree list`
Create worktree	`git worktree add ../{agent}-worktree -b {agent}-worktree`
Remove worktree	`git worktree remove ../{agent}-worktree`
Navigate to worktree	`cd ../{agent}-worktree`
Merge main	`git fetch origin main && git merge origin/main`
Push worktree	`git push origin {agent}-worktree`
Check current branch	`git branch --show-current`

Worktree Naming Convention

MUST follow this pattern: {agent}-worktree

Agent	Worktree Name	Worktree Path	Branch Name
Developer	`developer-worktree`	`../developer-worktree`	`developer-worktree`
Tech Artist	`techartist-worktree`	`../techartist-worktree`	`techartist-worktree`

Initial Setup (One-Time Per Agent)

bash

# Check if worktree already exists
git worktree list

# If NOT in list, create worktree with dedicated branch
git worktree add ../{agent}-worktree -b {agent}-worktree

# Verify creation
git worktree list

Daily Workflow

Before Starting ANY Task

⚠️ CRITICAL: You MUST navigate to your worktree BEFORE doing any development work!

Why navigate to worktree first?

•All code/asset edits MUST happen in worktree
•Commits go to worktree branch, NOT main
•QA validates from worktree, then merges to main
•Prevents file conflicts between parallel agents

Why merge main first?

•Ensures you have the latest changes from the other agent's work
•Prevents out-of-date code conflicts
•Both agents' work gets integrated through main

Unit, E2E, and Playwright MCP Testing Protocol

Remember to always run the tests in the target worktree branches for validation

•Move to the target worktree
•Load the correct skills for this scope
•Run validation process
•During validation process, always double check in which port the client is running for proper URL check
•Investigate in case of issues and iterate until fixed

•
Developer in worktree: ├── Reads PRD from master (Get-MasterPrdPath) ├── Reads messages from master (Get-MasterMessageQueuePath) ├── Updates PRD status in master (atomic write) ├── Sends messages to master queue ├── Writes code in worktree (src/) ├── Commits code to worktree branch └── Sends validation_request via master queue
•
PM in master: ├── Reads PRD (sees worker status updates) ├── Reads messages (sees worker messages) └── Coordinates everything
•
QA in master: ├── Navigates to worktree for testing ├── If validation passes: merges to master └── If validation fails: sends bug_report via master queue