/init

---

🚨 EXECUTION NOTICE FOR CLAUDE

When you invoke this command via SlashCommand, the system returns THESE INSTRUCTIONS below.

YOU are the executor. This is NOT an autonomous subprocess.

• ✅ The phases below are YOUR execution checklist
• ✅ YOU must run each phase immediately using tools (Bash, Read, Write, Edit, TodoWrite)
• ✅ Complete ALL phases before considering this command done
• ❌ DON't wait for "the command to complete" - YOU complete it by executing the phases
• ❌ DON't treat this as status output - it IS your instruction set

Immediately after SlashCommand returns, start executing Phase 0, then Phase 1, etc.

See @CLAUDE.md section "SlashCommand Execution - YOU Are The Executor" for detailed explanation.

---

Arguments: $ARGUMENTS

Goal

Create isolated git worktree for each spec, enabling parallel development without conflicts. All agents working on a spec share the same worktree.

Modes:

• <spec-name> - Single spec (e.g., 001-red-seal-ai)
• --all or --bulk - All specs at once (100+ specs supported)

Performance Optimization: Use pnpm for Multi-Worktree Development

Recommended for projects creating multiple worktrees:

If your project uses Node.js/npm, consider converting to pnpm for dramatic performance improvements:

# Convert your project (one-time setup)
/foundation:use-pnpm

# Or manually:
npm install -g pnpm
cd /path/to/your/project
rm -rf node_modules package-lock.json
pnpm install  # Creates pnpm-lock.yaml

Benefits for Worktree Workflows:

• ⚡ 80-90% faster installs - 5-10s instead of 30-60s per worktree
• 💾 75% less disk space - 600MB instead of 2GB for 4 worktrees
• 🔒 Still isolated - Each worktree gets correct dependency versions
• 🤖 Auto-detected - Worktree system automatically uses pnpm if available

How it works:

• First worktree: Downloads to global cache (~/.pnpm-store/)
• Subsequent worktrees: Create hardlinks from cache (near-instant)
• Each worktree still has isolated node_modules/ with correct versions

System automatically detects and uses pnpm - no changes to workflow needed!

Phase 1: Discovery

Actions:

• Check if bulk mode requested (--all or --bulk in arguments)

Single Spec Mode

• Parse spec name from arguments: $ARGUMENTS (e.g., F001 or 001-red-seal-ai)
• Find spec directory (phase-nested first, then legacy): !{bash find specs/phase-* -type d -name "$ARGUMENTS" 2>/dev/null | head -1 || find specs -type d -name "$ARGUMENTS" 2>/dev/null | head -1}
• Verify spec directory exists
• Extract spec number (e.g., "001" from "F001-red-seal-ai")
• Extract spec name (e.g., "red-seal-ai" from "F001-red-seal-ai")
• Get current git branch and verify clean working directory

Bulk Mode (--all or --bulk)

• Scan all specs across all phases: specs/phase-*/F*
• Fall back to legacy: specs/*/
• Extract spec numbers and names
• Count total worktrees to create (one per spec)
• Show summary and confirm with user

Phase 2: Worktree Creation & Mem0 Registration

Single Spec Mode

Actions:

• Invoke worktree-coordinator agent to:
  • Create single git worktree for the spec
  • Install dependencies in worktree (Node.js, Python)
  • Register worktree in Mem0 for global tracking
• The agent handles:
  • Extract spec number (e.g., "001" from "001-red-seal-ai")
  • Extract spec name (e.g., "red-seal-ai")
  • Create branch: spec-{spec-number}
  • Create worktree: git worktree add ../{project}-{spec-number} -b spec-{spec-number}
  • Register in Mem0: register-worktree.py register --spec {spec} --spec-name {name} --path {path} --branch {branch}
  • Install dependencies: register-worktree.py setup-deps --path {path} (auto-detects npm/pnpm/yarn/pip)
  • Display created worktree with git worktree list

Bulk Mode

Actions:

• Run bulk creation script:

  python plugins/planning/skills/doc-sync/scripts/bulk-register-worktrees.py
  

• Creates worktrees in parallel for all specs (10 concurrent)
• Registers all worktrees in Mem0 automatically
• Shows progress for each spec/agent combination
• Displays summary at end

Phase 3: Verification

Actions:

• Verify all worktrees created successfully
• Check each worktree is on correct branch
• Ensure layered-tasks.md symlinks are valid
• Count total worktrees created

Phase 4: Mem0 Verification

Actions:

• Query Mem0 to verify all registrations successful
• Run: register-worktree.py list to show active worktrees
• Confirm agents can query their assignments

Phase 5: Summary

Display:

• Spec name and number
• Worktree path and branch
• Dependencies installation status
• Registered in Mem0 for agent discovery
• How agents query: register-worktree.py get-worktree --spec {number}
• Next steps: Any agent working on this spec will automatically use the worktree