A production-tested template for organizing Claude Code into a multi-domain workspace with persistent memory, session continuity, daily task tracking, and intelligent onboarding.

Built by Tuning Labs — battle-tested across 30+ agent swarms, 70+ tracked items, and months of daily use building a vertical SaaS platform.

A ready-to-use template that transforms Claude Code from a simple CLI into an organized command center for your project. It includes:

• Interactive setup interview (/setup) that configures everything from your answers
• Session persistence — never lose context between sessions
• Daily task tracking — persistent open items list updated every session
• Multi-domain routing — separate contexts for coding, marketing, financial, support, ops
• Memory system — structured knowledge base that grows with your project
• Hook scripts — auto-save state on exit, auto-load on start, pre-commit safety checks
• Lessons learned — perpetual log that prevents repeating mistakes

# 1. Clone or copy this template into your project
git clone https://github.com/YOUR_USERNAME/claude-command-center.git
cp -r claude-command-center/.claude your-project/.claude
cp claude-command-center/CLAUDE.md your-project/CLAUDE.md

# 2. Start Claude Code in your project
cd your-project
claude

# 3. Run the setup interview
/setup

The setup interview asks about your project, tech stack, preferences, and workflow in 5 stages. It generates all configuration files automatically.

.claude/
  hooks/
    session-start.sh    # Loads prior session state on every start
    session-end.sh      # Auto-saves branch, commits, work on every exit
    pre-commit.sh       # Catches brand name errors, hardcoded secrets
  settings.local.json   # Hook configuration and permissions
  skills/               # Custom slash commands
  plans/                # Work-in-progress implementation plans

memory/
  MEMORY.md             # Index of all memory files
  templates/            # Blank templates for each memory type
  examples/             # Filled examples showing best practices

domains/
  coding/CLAUDE.md      # Development-specific context and rules
  marketing/CLAUDE.md   # Marketing-specific context and rules
  financial/CLAUDE.md   # Financial analysis context and rules
  support/CLAUDE.md     # Customer support context and rules
  operations/CLAUDE.md  # Operations and admin context and rules

CLAUDE.md               # Master router — project-wide rules and navigation

Every time you exit Claude Code, the session-end.sh hook captures your git branch, recent commits, uncommitted changes, and worktree status into session-closing.md. When you start a new session, session-start.sh loads that state back. You never lose context.

Memory files use YAML frontmatter with 4 types: user, feedback, project, reference. The MEMORY.md index keeps everything discoverable. Memory grows organically as you work — feedback from mistakes becomes guidance for future sessions.

The open-items-daily.md file persists across sessions. Items are organized by priority (Critical Path, Features, Research, Cleanup). Mark items complete as you go. Set up a daily reminder to review open items each morning.

Each domains/ folder has its own CLAUDE.md with domain-specific rules, tools, and conventions. When you cd into a domain folder and run Claude, it loads that domain's context automatically.

The pre-commit.sh hook checks for brand name misspellings, hardcoded secrets, forbidden file patterns, and API keys before every commit.

Create .md files in .claude/skills/ with YAML frontmatter:

---
name: my-skill
description: What this skill does
user_invocable: true
---
[Skill instructions here]

Edit .claude/hooks/pre-commit.sh — add patterns to BRAND_RULES, FORBIDDEN_PATTERNS, or FORBIDDEN_FILES arrays.

Copy a template from memory/templates/, fill it in, and add a link to memory/MEMORY.md.

This system is built on three principles:

1. Context is everything. The more Claude knows about your project, preferences, and history, the better it performs. Memory makes this persistent.

2. Never repeat mistakes. The lessons-learned file and feedback memories ensure that hard-won knowledge survives across sessions.

3. Track everything, complete everything. The open items tracker prevents work from falling through cracks across sessions. No partial work.

MIT License - Copyright (c) 2026 Tuning Labs (tuninglabs.net)

See LICENSE for details.