Tool Stack: Follow ~/.claude/skills/_shared/tool-stack.md for ALL tool choices.
Quality Gates: Follow ~/.claude/skills/_shared/quality-gates.md for ALL quality checks.

Claude Code Starter Pack

The complete, up-to-date system that turns Claude Code from "AI chatbot" into an Agent Operating System. Give this to ANYONE and they get the most efficient Claude Code experience that exists.

The Luis Carrillo Promise

You don't need to be techie. You don't need to code. You just talk.

The single biggest reason people don't use AI tools that could change their lives is FEAR. Fear of breaking something. Fear of looking stupid. Fear of the terminal. Fear of not understanding what's happening.

That fear is a lie. And this starter pack is the antidote.

Claude Code is a conversation. You type what you want. It does the work. The hooks, the configs, the MCP servers — all of that is a one-time setup that turns a basic AI into YOUR AI. After that? You just talk.

Luis Carrillo built this because he watched brilliant, creative people stop before they started. Not because they couldn't do it. Because nobody told them they didn't need to be technical to use the most powerful productivity tool that has ever existed.

This is that conversation.

---

Why This Exists

Most people use Claude Code at 10% capacity. They type prompts, click approve, lose context when sessions get long, and never learn that Claude Code has 14 hook lifecycle events, 3 hook types, agent teams, worktree isolation, 1M token context, and a full plugin ecosystem.

This starter pack is the difference between struggling and shipping.

What You Get

Component	What It Does	Impact
11 Custom Hooks	Context survival, smart boot, cost tracking, session persistence	Session never dies. Context never lost.
CLAUDE.md Template	Cognitive OS, visual-first output, gate protocol, execution modes	Claude thinks like YOU, not like a generic bot
Features Index	Queryable reference of ALL Claude Code capabilities	Never wonder "can Claude do X?" again
Daily Update Monitor	Auto-checks for new Claude Code versions and features	Never be outdated. Ever.
14+ MCP Servers	Gmail, Notion, Drive, EXA, Perplexity, Brave, Gemini, Tavily, etc.	Claude can DO things, not just talk
184+ Skills	Content, video, cloning, coaching, engineering, automation	Expertise on demand

---

Quick Start (New Machine / New Person)

Step 1: Install Claude Code

npm install -g @anthropic-ai/claude-code

Step 2: Create Directory Structure

mkdir -p ~/.claude/{hooks,skills,references,scripts,agents,commands,memory}
mkdir -p ~/.claude/projects

Step 3: Install Essential Hooks

These are the hooks that make Claude Code actually usable:

Hook 1: Context Survival (THE breakthrough)

What it does: Before context compaction, saves session state to a file. After compaction, reads it back. Your chat literally never dies.

Event: PreCompact
File: ~/.claude/hooks/context-survival.js

This is the single most important hook. Without it, long sessions lose context and you start over. With it, sessions are infinite.

Hook 2: Smart Boot

What it does: At session start, detects your project, loads relevant context, checks for deferred tasks, surfaces recurring frustrations, and alerts you to Claude Code updates.

Event: SessionStart
File: ~/.claude/hooks/smart-boot.js

Hook 3: Session End Enforcer

What it does: When a session ends, forces saves to supermemory, MEMORY.md, session-intelligence, and deferred-tasks. Nothing gets lost.

Event: Stop
File: ~/.claude/hooks/session-end-enforcer.js

Hook 4: Cost Guardian

What it does: Monitors API spend in real-time. Alerts at thresholds. Prevents surprise bills.

Event: PostToolUse (matcher: *)
File: ~/.claude/hooks/cost-guardian.js

Hook 5: Tool Routing Guard

What it does: Ensures Claude uses the right tools — Read instead of cat, Grep instead of grep, Edit instead of sed. Prevents bad habits.

Event: PreToolUse (matcher: Bash)
File: ~/.claude/hooks/tool-routing-guard.js

Step 4: Configure settings.json

{
  "hooks": {
    "SessionStart": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "node \"$HOME/.claude/hooks/smart-boot.js\""
          }
        ]
      }
    ],
    "PreCompact": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "node \"$HOME/.claude/hooks/context-survival.js\""
          }
        ]
      }
    ],
    "Stop": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "node \"$HOME/.claude/hooks/session-end-enforcer.js\""
          }
        ]
      }
    ],
    "PreToolUse": [
      {
        "matcher": "Bash",
        "hooks": [
          {
            "type": "command",
            "command": "node \"$HOME/.claude/hooks/tool-routing-guard.js\""
          }
        ]
      }
    ],
    "PostToolUse": [
      {
        "matcher": "*",
        "hooks": [
          {
            "type": "command",
            "command": "node \"$HOME/.claude/hooks/cost-guardian.js\""
          }
        ]
      }
    ]
  }
}

Step 5: Install Daily Update Monitor

# Copy the monitor script
cp ~/.claude/scripts/claude-update-monitor.sh ~/.claude/scripts/

# Make executable
chmod +x ~/.claude/scripts/claude-update-monitor.sh

# Install launchd daemon (runs daily at 6 AM)
cp ~/Library/LaunchAgents/com.luis.claude-update-monitor.plist ~/Library/LaunchAgents/
launchctl load ~/Library/LaunchAgents/com.luis.claude-update-monitor.plist

# Run immediately
bash ~/.claude/scripts/claude-update-monitor.sh

Step 6: Create CLAUDE.md

Your CLAUDE.md is where you tell Claude WHO YOU ARE and HOW to work. At minimum:

# Project Instructions

## Communication Style
- Visual-first: tables, box frames, checklists
- Max 3 sentences before a visual break
- Scannable in 5 seconds

## Execution Rules
- Read files before editing them
- Use dedicated tools (Read, Edit, Write) not bash
- Parallel tool calls when independent
- Never suggest "new session" — sessions are infinite

## What I'm Building
[Your project description here]

---

The 14 Hook Lifecycle Events

#	Event	Fires When	Can Block?	Essential?
1	SessionStart	Session begins	NO	✅ YES
2	UserPromptSubmit	User hits enter	YES	Optional
3	PreToolUse	Before tool runs	YES	✅ YES
4	PermissionRequest	Permission dialog	YES	Optional
5	PostToolUse	After tool succeeds	NO	✅ YES
6	PostToolUseFailure	After tool fails	NO	Optional
7	SubagentStop	Sub-agent completes	NO	Optional
8	Stop	Response ends	NO	✅ YES
9	Notification	Claude notifies	NO	Optional
10	PreCompact	Before compaction	NO	✅ YES
11	ConfigChange	Config file changes	YES	Optional
12	statusLine	Status bar update	N/A	Nice to have
13	WorktreeCreate	Worktree created	N/A	Optional
14	WorktreeRemove	Worktree removed	N/A	Optional

3 Hook Types

Type	How It Works	Best For
command	Runs shell command	Scripts, formatting, logging
prompt	Uses LLM evaluation	Smart validation, context-aware decisions
agent	Spawns a sub-agent	Complex multi-step automation
HTTP	POSTs JSON to URL	External service integration

Hook JSON Structure

{
  "hooks": {
    "EventName": [
      {
        "matcher": "ToolPattern",
        "hooks": [
          {
            "type": "command",
            "command": "node script.js",
            "timeout": 60
          }
        ]
      }
    ]
  }
}

Exit Codes

Code	Meaning	Behavior
0	Success	Continue normally
2	Block	Stop action, show error
Other	Error	Log only, continue

---

Agent Capabilities

Approach	What	When to Use
Sub-agents (Task tool)	Parallel isolated work	Research, multi-file analysis
.claude/agents/*.md	Persistent specialist agents	Recurring roles (reviewer, tester)
Agent Teams	Multi-agent coordination	Complex projects
Worktree Isolation	-w flag, separate git branch	Parallel branches
Background Agents	background: true	Non-blocking work

---

Essential Slash Commands

Command	What It Does
/compact	Force context compaction
/resume	Resume previous conversation
/model	Switch models
/memory	Manage auto-memory
/cost	Check API spend
/simplify	Simplify code (NEW)
/batch	Batch operations (NEW)
/fork	Fork conversation

---

MCP Servers Worth Installing

Server	What It Gives You
Gmail	Read, search, draft, send emails
Google Calendar	View, create, manage events
Notion	Full database/page access
Google Drive	File management
EXA	Web search with clean content
Perplexity	AI-powered web search
Brave Search	Privacy-focused web search
Tavily	Research-grade web search
Gemini Grounding	Google search + Reddit
Supermemory	Cross-session memory
Context7	Library documentation lookup
Scrapling (advanced)	Adaptive web scraping — bypasses bot detection, Cloudflare

---

Daily Update Monitor

A launchd daemon that runs every day at 6 AM:

• Checks your Claude Code version against latest
• Writes update notices for smart-boot to surface
• Tracks ecosystem changes (compound-engineering, skills)
• Updates the features index reference file

Files:

• Script: ~/.claude/scripts/claude-update-monitor.sh
• LaunchAgent: ~/Library/LaunchAgents/com.luis.claude-update-monitor.plist
• Features Index: ~/.claude/references/claude-code-features-2026.md
• Update Notice: ~/.claude/references/.update-notice

---

The Philosophy

Claude Code is not a chatbot. It's an Agent Operating System. The hooks make it deterministic. The skills make it specialized. The CLAUDE.md makes it YOU.

The difference between someone who "uses Claude Code" and someone who SHIPS with Claude Code is exactly this setup. Every hook, every skill, every config — it's the difference between a bicycle and a motorcycle.

The starter pack is the motorcycle.

---

Maintenance

• Daily: Update monitor runs automatically (6 AM)
• Weekly: Check ~/.claude/references/claude-code-features-2026.md for new features
• Monthly: Audit hooks for redundancy, update CLAUDE.md
• When notified: Run npm update -g @anthropic-ai/claude-code

Essential Fix: Large Input Protocol (50K Paste Limit)

Claude Code truncates user messages at ~50,000 characters — no warning, no error, content just disappears. This kills transcripts, long code dumps, and anything over ~12 pages.

The fix: Save clipboard to a temp file, tell Claude to read it.

One-Time Setup

# Create the script
cat > ~/.claude/scripts/save-clipboard.sh << 'SCRIPT'
#!/bin/bash
NAME="${1:-$(date +%Y%m%d-%H%M%S)}"
OUTPUT="/tmp/claude-large-input-${NAME}.txt"
pbpaste > "$OUTPUT"
CHARS=$(wc -c < "$OUTPUT")
LINES=$(wc -l < "$OUTPUT")
echo "╔══════════════════════════════════════════════╗"
echo "║  SAVED: $OUTPUT"
echo "║  Size: ${CHARS} chars, ${LINES} lines"
echo "║  Tell Claude: read /tmp/claude-large-input-${NAME}.txt"
echo "╚══════════════════════════════════════════════╝"
SCRIPT
chmod +x ~/.claude/scripts/save-clipboard.sh

# Add alias
echo 'alias sc="bash ~/.claude/scripts/save-clipboard.sh"' >> ~/.zshrc
source ~/.zshrc

How to Use

1. Copy your large content (Cmd+C)
2. In a separate terminal: sc transcript (or any name)
3. In Claude Code: "read /tmp/claude-large-input-transcript.txt"

Full skill: ~/.claude/skills/large-input/SKILL.md

---

Reference Files

• Features Index: ~/.claude/references/claude-code-features-2026.md
• Hook Reference: ~/.claude/skills/building-hooks/references/
• SSH Patterns: ~/.claude/skills/ssh-patterns/SKILL.md
• Mac Automation: ~/.claude/skills/mac-automation/SKILL.md
• Large Input: ~/.claude/skills/large-input/SKILL.md

---

Level 2: Multi-Device Bridge (Optional)

For users with 2+ Macs. Turns separate machines into one unified Claude Code system — auto-syncing skills, hooks, scripts, references, and configs between devices. Zero manual commands. Zero maintenance.

What You Get

Feature	What It Does
Two-way SSH bridge	Passwordless SSH between all your Macs using ed25519 keys
Auto-sync LaunchAgent	Every 30 min, skills/hooks/scripts/references sync automatically
SSH config pinning	No "too many auth failures" — keys locked to one identity
Natural language control	"Do X on my iMac" → Claude uses the bridge silently
Zero maintenance	Once set up, never think about it again

Prerequisites

• 2+ Macs on the same network (or reachable via hostname/IP)
• Remote Login enabled on each machine (System Settings → General → Sharing → Remote Login)
• Claude Code installed on each machine

Setup: Multi-Device Bridge

Step 1: Generate SSH Keys (each machine)

Run on EACH machine that doesn't already have an ed25519 key:

# Check if key exists
ls ~/.ssh/id_ed25519.pub 2>/dev/null || ssh-keygen -t ed25519 -C "$(whoami)@$(hostname -s)" -N "" -f ~/.ssh/id_ed25519

Step 2: Exchange Keys

Copy each machine's public key to every OTHER machine's ~/.ssh/authorized_keys:

# From Machine A — add Machine A's key to Machine B
ssh-copy-id -i ~/.ssh/id_ed25519.pub user@machine-b.local

# From Machine B — add Machine B's key to Machine A
ssh-copy-id -i ~/.ssh/id_ed25519.pub user@machine-a.local

If ssh-copy-id fails, do it manually:

# On Machine A, get the key
cat ~/.ssh/id_ed25519.pub
# Paste that line into Machine B's ~/.ssh/authorized_keys

Step 3: Create SSH Config (each machine)

Create ~/.ssh/config on EACH machine pointing to every other machine:

# ~/.ssh/config on Machine A
Host machine-b machine-b.local
    HostName machine-b.local
    User yourusername
    IdentityFile ~/.ssh/id_ed25519
    IdentitiesOnly yes

# ~/.ssh/config on Machine B
Host machine-a machine-a.local
    HostName machine-a.local
    User yourusername
    IdentityFile ~/.ssh/id_ed25519
    IdentitiesOnly yes

Why IdentitiesOnly yes: Without this, SSH tries every key in your agent, hits "too many authentication failures" after 6 attempts, and locks you out. This pins to ONE key.

Step 4: Verify Round-Trip

# From Machine A
ssh machine-b "echo 'Machine B reached from A'"

# From Machine B
ssh machine-a "echo 'Machine A reached from B'"

Both must succeed without a password prompt.

Step 5: Create Auto-Sync Script

Create ~/.claude/scripts/multi-device-sync.sh on EACH machine:

#!/bin/bash
# Multi-Device Sync — syncs Claude Code config to other machines
# NEVER uses --delete (safety first)

LOG="$HOME/.claude/logs/sync-$(date +%Y%m%d).log"
mkdir -p "$HOME/.claude/logs"
log() { echo "[$(date '+%H:%M:%S')] $1" >> "$LOG"; }

# === CONFIGURE THESE ===
REMOTE_HOST="machine-b.local"  # Change to your other machine
REMOTE_USER="yourusername"      # Change to your username
REMOTE_SSH="ssh -o IdentitiesOnly=yes -i $HOME/.ssh/id_ed25519"
# ========================

log "=== SYNC START ==="

if ! ping -c 1 -W 2 "$REMOTE_HOST" &>/dev/null; then
  log "$REMOTE_HOST not reachable — skipping"
  exit 0
fi

# Sync Claude Code directories (NEVER --delete)
for dir in hooks scripts references skills/configs; do
  if [ -d "$HOME/.claude/$dir" ]; then
    rsync -az -e "$REMOTE_SSH" "$HOME/.claude/$dir/" "$REMOTE_USER@$REMOTE_HOST:~/.claude/$dir/" 2>/dev/null
    log "Synced $dir"
  fi
done

# Sync individual config files
for file in settings.json CLAUDE.md; do
  if [ -f "$HOME/.claude/$file" ]; then
    rsync -az -e "$REMOTE_SSH" "$HOME/.claude/$file" "$REMOTE_USER@$REMOTE_HOST:~/.claude/$file" 2>/dev/null
    log "Synced $file"
  fi
done

log "=== SYNC COMPLETE ==="

chmod +x ~/.claude/scripts/multi-device-sync.sh

Step 6: Install Auto-Sync LaunchAgent

Create ~/Library/LaunchAgents/com.claude.multi-device-sync.plist:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>Label</key>
    <string>com.claude.multi-device-sync</string>
    <key>ProgramArguments</key>
    <array>
        <string>/bin/bash</string>
        <string>~/.claude/scripts/multi-device-sync.sh</string>
    </array>
    <key>StartInterval</key>
    <integer>1800</integer>
    <key>RunAtLoad</key>
    <true/>
    <key>StandardErrorPath</key>
    <string>/tmp/claude-sync-err.log</string>
    <key>StandardOutPath</key>
    <string>/tmp/claude-sync-out.log</string>
    <key>EnvironmentVariables</key>
    <dict>
        <key>PATH</key>
        <string>/opt/homebrew/bin:/usr/local/bin:/usr/bin:/bin</string>
    </dict>
</dict>
</plist>

launchctl load ~/Library/LaunchAgents/com.claude.multi-device-sync.plist

Do this on EACH machine, pointed at the other machine.

Step 7: Add to CLAUDE.md

Add this to your CLAUDE.md so Claude knows about your other machines:

## Multi-Device Bridge

Machines on my network:
- Machine A: machine-a.local (primary)
- Machine B: machine-b.local (secondary)

SSH bridge is configured. "Do X on my [machine name]" = Claude uses SSH silently. Never show SSH commands. Never ask me to run terminal commands on the other machine. Just do it.

What Syncs Automatically

Directory/File	Syncs?	Why
~/.claude/hooks/	YES	Same hooks on all machines
~/.claude/scripts/	YES	Same automation everywhere
~/.claude/references/	YES	Same knowledge base
~/.claude/skills/configs/	YES	Same skill config
~/.claude/settings.json	YES	Same hook wiring
~/.claude/CLAUDE.md	YES	Same personality/rules
~/.claude/skills/ (full)	NO*	Use git repo for skills (too large for rsync)

*Skills should be managed via a git repo (git pull on each machine) for version control. The sync handles everything else.

Troubleshooting

Problem	Fix
"Too many authentication failures"	Add IdentitiesOnly yes to ~/.ssh/config
"Connection refused"	Enable Remote Login in System Settings
"Host key verification failed"	Run ssh -o StrictHostKeyChecking=accept-new machine.local once
Sync not running	Check: `launchctl list
Wrong hostname	Run hostname on target machine to verify

The 3 Levels

Level	What You Get	Who It's For
Level 1: Solo Machine	Hooks, CLAUDE.md, skills, MCP servers	Everyone starting out
Level 2: Multi-Device Bridge	+ SSH bridge, auto-sync, natural language device control	Power users with 2+ Macs
Level 3: Full Agent OS	+ 185 skills, Cognitive OS, clone system, automated backups to GitHub + Google Drive	Full-time builders

---

Built by Agent Luis. Updated automatically. Never outdated.

Gate Protocol

• Pre-flight: Verify all dependencies and tools are available. Check file paths exist. Confirm no destructive operations without backup.
• Mid-flight: Log each operation. Verify intermediate outputs before proceeding to next step. Handle errors gracefully — never fail silently.
• Post-flight: Verify system state is correct. Confirm all outputs saved to expected locations. Test that the system works end-to-end.

Change Log

• [2026-03-02]: War Room Tier 2 — content quality upgrade (gate→system).
• [2026-03-02]: War Room Tier 1 audit — skill structure verified and standardized.