The Complete AI Development Operating System for Claude Code Transform Claude Code from a coding assistant into a full engineering team.
A production-ready CLAUDE.md system that solves the biggest problems developers face with Claude Code on large projects:
| Problem | How This Kit Solves It |
|---|---|
| Claude Code skips details on multiple tasks | Deep Focus Mode — single-task enforcement |
| Features are static/don't work | Zero Mock Data Rule — real API + DB connections enforced |
| Testing is superficial | Action-Level Testing — every button clicked, every form submitted |
| No self-review after implementation | Mandatory Self-Review — 7-category rating, 8.5/10 threshold |
| Context lost when starting new chat | MemPalace Integration — persistent memory across sessions |
| Claude Code doesn't give 100% effort | Max Effort Protocol — lazy output detection + quality gates |
| Session ends without validation | Session Exit Gate — codebase review before summary |
| Design quality poor vs Lovable.dev | Design System — DESIGN.md + shadcn/ui + ShadcnBlocks |
| Browser testing burns too many tokens | Playwright CLI — 4x more efficient than Playwright MCP |
| Multi-project context bleed | MemPalace Wings — per-project isolated memory |
nexalance-claude-code-kit/
├── NexaLance-CLAUDE-v4.3-SUPREME.md ← Master CLAUDE.md template (2,350+ lines)
├── setup-nexalance.sh ← One-time machine setup script
├── setup-project-wing.sh ← Per-project setup script
├── README.md ← This file
└── LICENSE ← MIT License
git clone https://github.com/developerjillur/nexalance-claude-code-kit.git ~/Desktop/nexalance-kit
cd ~/Desktop/nexalance-kit
chmod +x setup-nexalance.sh setup-project-wing.shbash setup-nexalance.shThis installs: Superpowers, MemPalace, Code Review, Feature Dev, Frontend Design, Playwright CLI, Chrome DevTools MCP, Vercel Design Guidelines, and auto-save hooks.
mkdir -p ~/projects/my-project && cd ~/projects/my-project
# Copy the master CLAUDE.md
cp ~/Desktop/nexalance-kit/NexaLance-CLAUDE-v4.3-SUPREME.md ./CLAUDE.md
# Initialize project with isolated memory wing
bash ~/Desktop/nexalance-kit/setup-project-wing.sh "my-project" "My Awesome Project" "client-name"
# Add your PRD
# Put your Product Requirements Document in docs/PRD.md
# Open in VS Code
code .Open Claude Code panel in VS Code → type:
Start project
Claude Code may ask: "What would you like to start?" or "Is this a new project?"
Reply with this:
This is a new project. Read the CLAUDE.md in the project root and follow the Phase 0 initialization protocol. The PRD is in docs/PRD.md which references the PRD files in docs/ folder. Read all of them and generate all 9 documents as instructed in CLAUDE.md.
After that, Claude Code will automatically:
- Read your CLAUDE.md (2,350+ lines of instructions)
- Read your PRD file(s)
- Generate 9 documents (DESIGN, TAD, SCHEMA, API, RULES, FEATURES, TASKS, TESTING, SESSION)
- Print summary
- Ask: "Ready for Phase 1. Say 'Go' to start building."
- Type "Go" → development starts!
| Tool | Required | Check Command |
|---|---|---|
| Node.js 18+ | ✅ | node --version |
| Python 3.9+ | ✅ | python3 --version |
| Git | ✅ | git --version |
| Claude Code CLI | ✅ | claude --version |
| VS Code | Recommended | - |
cd ~/Desktop/nexalance-kit
bash setup-nexalance.shWhat this script does:
✅ Checks prerequisites (Python, Node, Git, Claude CLI)
✅ Installs MemPalace (persistent AI memory)
✅ Installs Superpowers plugin (brainstorm + plan + TDD workflow)
✅ Installs Code Review plugin (5-agent parallel review)
✅ Installs Feature Dev plugin (7-phase implementation)
✅ Installs Frontend Design plugin (anti-AI-slop design)
✅ Installs Playwright CLI (4x token-efficient browser testing)
✅ Installs Chromium browser (for Playwright)
✅ Configures Chrome DevTools MCP (browser debugging)
✅ Configures MemPalace MCP (user-level, available in all projects)
✅ Configures auto-save hooks (every 15 messages + before /compact)
✅ Installs ShadcnBlocks skill (2,500+ UI blocks)
✅ Installs Vercel Web Design Guidelines (100+ UX rules)
cd ~/projects/your-project
# Copy master CLAUDE.md
cp ~/Desktop/nexalance-kit/NexaLance-CLAUDE-v4.3-SUPREME.md ./CLAUDE.md
# Run project setup (creates all config files + folders)
bash ~/Desktop/nexalance-kit/setup-project-wing.sh "wing-name" "Project Full Name" "client"Parameters:
| Parameter | Description | Example |
|---|---|---|
wing-name |
Short identifier (lowercase, no spaces) | my-saas-app |
Project Full Name |
Human-readable name | My SaaS Application |
client |
Client identifier (optional) | client123 |
What this script creates:
your-project/
├── CLAUDE.md ← With PROJECT_WING configured
├── .mcp.json ← MemPalace + TaskMaster servers
├── .gitignore ← Proper ignore rules
├── .claude/
│ ├── settings.json ← Auto-save hooks
│ └── commands/
│ └── auto-orchestrator.md ← Plugin auto-routing skill
├── .screenshots/
│ ├── tests/ ← E2E test screenshots
│ ├── reviews/ ← Design review screenshots
│ ├── deployments/ ← Deploy proof screenshots
│ └── bugs/ ← Bug evidence screenshots
├── docs/
│ ├── SESSION.md ← Session state tracker
│ ├── HANDOFF.md ← Session handoff for new chats
│ ├── reviews/ ← Code review reports
│ ├── analysis/ ← Technical analysis docs
│ └── plans/ ← Implementation plans
└── .git/ ← Git initialized
1. Open VS Code with your project folder
2. Open Claude Code panel (Spark icon in sidebar)
3. Type: "Continue" or "Resume"
4. Claude Code loads MemPalace memory → reads SESSION.md → resumes exact task
Note: If Claude Code asks what to do, reply:
"Read the CLAUDE.md and docs/SESSION.md, then resume where you left off."
1. Open different project folder in VS Code
2. Open Claude Code panel
3. Type: "Continue"
4. MemPalace automatically switches to that project's wing
5. Zero context bleed between projects
1. Type: "I'm done for today"
2. Claude Code automatically:
→ Runs Session Exit Gate (codebase review + tests)
→ Runs /code-review
→ Saves memory to MemPalace
→ Updates SESSION.md + HANDOFF.md
→ Git commits all progress
"Build the user auth system" → /feature-dev auto-runs
"Test the login form" → Playwright CLI opens browser
"Check why this page is broken" → Chrome DevTools checks console
"Review the code" → /code-review 5-agent analysis
"Plan the payment integration" → Superpowers brainstorm
"Deploy to hosting" → Playwright CLI headed browser
| Plugin | Purpose | Auto-triggers When |
|---|---|---|
| Superpowers | Brainstorm → Plan → TDD → Execute | New ideas, planning |
| MemPalace | Persistent memory across sessions | Every session start/end |
| Code Review | 5-agent parallel code review | After every feature |
| Feature Dev | 7-phase feature implementation | Building complex features |
| Frontend Design | Anti-AI-slop design quality | Any frontend work |
| Playwright CLI | Browser testing (4x efficient) | Testing, deploying |
| Chrome DevTools | Console, network, performance | Debugging browser issues |
| Vercel Guidelines | 100+ UX/accessibility rules | UI quality checks |
- Auto-Orchestration — All plugins trigger automatically based on context
- Deep Focus Mode — Single-task enforcement, no shallow multi-tasking
- Maximum Effort Protocol — Lazy output detection + quality enforcement
- Plugin Priority Rules — Clear conflict resolution between plugins
- Phase 0: Document Generation — 9 docs auto-generated from PRD (DESIGN, TAD, SCHEMA, API, RULES, FEATURES, TASKS, TESTING, SESSION)
- Phase-Based Development — 5 phases with integration checkpoints
- Feature Workflow — 8-step: context → backend → frontend → chain verify → review → git → memory → next
- Design System — DESIGN.md auto-generated with colors, typography, spacing, component tokens
- Action-Level Testing — Every button, form, CRUD operation tested via Playwright CLI
- Self-Review & Rating — 7-category scoring (including Design Quality), 8.5/10 threshold
- Session Exit Gate — 5-step validation before ending any session
- Security Checklist — Applied to every feature
- MemPalace Wings — Per-project isolated memory, zero cross-project bleed
- Context Management — Progressive disclosure, rate limit handling
- HANDOFF.md — Session context preservation for new chats
- Token Tracking — Context health monitoring, /compact suggestions
- 9 Common Patterns — Pagination, search, upload, soft delete, audit, WebSocket, email, cron, job queue
- Anti-Patterns — Code + workflow + architecture + 12 design anti-patterns
- Conflict Resolution — PRD contradictions, business rule conflicts, tech decisions
- Git Branch Workflow — Feature branches with merge protocol
- Organized File Output — Screenshots in
.screenshots/, reviews/analysis/plans indocs/
Each project gets its own MemPalace "wing" — completely isolated memory:
# Project 1
cd ~/projects/project-alpha
cp ~/Desktop/nexalance-kit/NexaLance-CLAUDE-v4.3-SUPREME.md ./CLAUDE.md
bash ~/Desktop/nexalance-kit/setup-project-wing.sh "alpha" "Project Alpha" "client-a"
# Project 2
cd ~/projects/project-beta
cp ~/Desktop/nexalance-kit/NexaLance-CLAUDE-v4.3-SUPREME.md ./CLAUDE.md
bash ~/Desktop/nexalance-kit/setup-project-wing.sh "beta" "Project Beta" "client-b"
# Switching: just open the folder → Claude Code loads correct wingMemPalace "not found" error
pip3 install mempalace --break-system-packages
python3 -m mempalace init ~/mempalace-dataPlaywright CLI "not found"
npm install -g @playwright/cli
npx playwright install chromiumPlugin not installed
Open Claude Code and run:
/plugin marketplace add obra/superpowers-marketplace
/plugin install superpowers@superpowers-marketplace
/plugin install code-review@claude-plugins-official
/plugin install feature-dev@claude-plugins-official
/plugin install frontend-design@claude-plugins-official
Context getting heavy / slow responses
Type in Claude Code: /compact
Or start a new session — MemPalace preserves all context.
Browser test needs login
Tell Claude Code: "Open browser and navigate to [URL]" Browser opens (headed mode) → you log in manually → tell Claude Code: "Continue testing"
New session lost context
Type: "Read docs/HANDOFF.md and docs/SESSION.md, then resume" With MemPalace installed, this happens automatically.
MCP servers show "failed" in project
If MemPalace shows "failed" at project level but "connected" at user level — that's fine. User-level takes priority. Remove duplicate from project .mcp.json:
echo '{}' > .mcp.json| Version | Lines | Score | Key Addition |
|---|---|---|---|
| v2.0 | 415 | 7.2/10 | Basic orchestrator |
| v3.0 | 913 | 9.1/10 | Security, deployment, rollback |
| v4.0 | 1,409 | 9.6/10 | Deep focus, action testing, self-review, memory |
| v4.1 | 1,641 | 9.6/10 | Superpowers + MemPalace + Official plugins |
| v4.2 | 1,793 | 8.4/10 | Playwright CLI + Chrome DevTools |
| v4.3 | 2,350+ | 9.2/10 | Design system, conflict resolution, file organization |
Built by NexaLance Agency — a full-service digital agency delivering 500+ projects across 30+ countries since 2016.
- Website: nexalance.com
- Email: info@sorobindu.com
MIT License — free to use, modify, and distribute.
NexaLance Agency — Your Success, Our Mission
