Claude Multi-PC Setup Guide

How to replicate Paul's full Claude environment — including all memory, sources, tools, MCP integrations, and workflows — across 7 Windows 11 machines on the Hot Springs and Texas networks.

Source machine: Paul's current Windows 11 Pro PC (walhu)
Claude Code version: 2.1.112
Date: April 17, 2026

Step 1: Install Claude Products

1A. Claude Code (CLI) — the terminal tool

This is the primary tool we've been using. It runs in your terminal (PowerShell, Git Bash, or Warp).

# Open PowerShell as Administrator and run:
npm install -g @anthropic-ai/claude-code

# If npm is not installed, install Node.js first:
winget install OpenJS.NodeJS.LTS

# Verify:
claude --version

1B. Claude Desktop (Windows app)

# Download from:
# https://claude.ai/download

# Or via winget:
winget install Anthropic.Claude

Claude Desktop gives you the chat interface with MCP server support (Gmail, Calendar, Slack).

1C. Claude Code in VS Code

# In VS Code, open Extensions (Ctrl+Shift+X)
# Search for "Claude Code"
# Install the official Anthropic extension
# Or from command line:
code --install-extension anthropic.claude-code

1D. Claude Code in JetBrains IDEs

# In any JetBrains IDE (IntelliJ, WebStorm, PyCharm):
# Settings → Plugins → Marketplace → search "Claude Code"
# Install and restart

1E. Claude Web (claude.ai/code)

No installation needed — just go to claude.ai/code in your browser and log in.

Step 2: Authenticate

# Run Claude Code and follow the login prompt:
claude

# It will open a browser for Anthropic account login
# Use your Anthropic account credentials
# The API key is stored automatically after login

Step 3: Copy Configuration & Memory

This is the critical step that makes the new machine "remember" everything.

3A. Copy the .claude directory

# On the SOURCE machine (walhu), create a zip:
powershell -c "Compress-Archive -Path 'C:\Users\walhu\.claude\*' -DestinationPath 'C:\Users\walhu\claude-config-backup.zip'"

# Transfer to the new machine via USB, network share, or cloud
# On the NEW machine, extract to the same path:
powershell -c "Expand-Archive -Path 'claude-config-backup.zip' -DestinationPath 'C:\Users\USERNAME\.claude\' -Force"

What this copies:

3B. Critical memory files that MUST transfer

These are in .claude/projects/C--Users-walhu/memory/:

MEMORY.md                          — Master index of all memories
barney-project-recovery.md         — READ FIRST after compaction (Barney project)
barney-factcheck-corrections.md    — Never repeat these errors
barney-autobiography-sources.md    — Where source files live
feedback_always-publish-live.md    — Push to droplet before reporting done
feedback_paulsjobs-on-blocked.md   — Add blocked URLs to Paul's Jobs page
feedback_be-proactive.md           — Anticipate problems
feedback_keep-building-people-database.md
feedback_always-push-everywhere.md — Push to droplet + B2 always
user_profile.md                    — Who Paul is
user_situation.md                  — Homes, financial context
+ many more project/reference memories

Step 4: Install Required Tools

# Python (for b2, yt-dlp, push_watcher)
winget install Python.Python.3.14

# Git (for version control)
winget install Git.Git

# VS Code
winget install Microsoft.VisualStudioCode

# yt-dlp (video downloader)
pip install yt-dlp

# Backblaze B2 CLI
pip install b2

# rclone (cloud sync)
winget install Rclone.Rclone

Step 5: Set Up SSH Keys for Droplet Access

# Generate SSH key (if new machine):
ssh-keygen -t ed25519 -C "paul@newmachine"

# Copy public key to droplet:
type C:\Users\USERNAME\.ssh\id_ed25519.pub | ssh root@143.198.182.180 "cat >> ~/.ssh/authorized_keys"

# Test:
ssh root@143.198.182.180 "echo 'Connected'"

Step 6: Set Up the Source Archive Access

Every machine needs access to the Barney Ebsworth source files (115 files, 355MB).

Option A: Pull from droplet (recommended)

# Create the local directory:
mkdir C:\Users\USERNAME\websites\barneyebsworth.com\sources

# Pull everything from the droplet:
scp -r root@143.198.182.180:/var/www/barneyebsworth.com/sources/* C:\Users\USERNAME\websites\barneyebsworth.com\sources\

Option B: Pull from B2

# Authorize b2:
C:\Users\USERNAME\AppData\Local\Python\...\Scripts\b2.exe authorize-account YOUR_KEY_ID YOUR_APP_KEY

# Sync from B2:
C:\Users\USERNAME\AppData\Local\Python\...\Scripts\b2.exe sync b2://walhus-private/barneyebsworth-sources/ C:\Users\USERNAME\websites\barneyebsworth.com\sources\

Option C: Copy from walhu via network share

# If walhu is on the same network:
xcopy \\WALHU\Users\walhu\websites\barneyebsworth.com\sources C:\Users\USERNAME\websites\barneyebsworth.com\sources /E /I

Step 7: Set Up the Push Watcher (auto-sync)

# Copy the watcher scripts:
mkdir C:\Users\USERNAME\websites\.sync
scp root@143.198.182.180:/var/www/barneyebsworth.com/../.sync/push_watcher.py C:\Users\USERNAME\websites\.sync\

# Or copy from walhu:
copy C:\Users\walhu\websites\.sync\push_watcher.py C:\Users\USERNAME\websites\.sync\

# Create a VBS launcher in Startup folder:
# (shell:startup → paste this as push_watcher.vbs)

' push_watcher.vbs — put in shell:startup
Set WshShell = CreateObject("WScript.Shell")
WshShell.Run "python C:\Users\USERNAME\websites\.sync\push_watcher.py", 0, False

Step 8: Configure MCP Servers

MCP (Model Context Protocol) servers give Claude access to Gmail, Google Calendar, Slack, etc.

# MCP servers are configured in Claude Desktop settings
# File: %APPDATA%\Claude\claude_desktop_config.json
# Copy this file from walhu to get the same MCP setup

Step 9: Set Up rclone Remotes

# Configure droplet remote:
rclone config
# → New remote → name: droplet → type: sftp
# → host: 143.198.182.180 → user: root → key_file: path to ssh key

# Configure NAS remote (if on Hot Springs network):
rclone config
# → New remote → name: cc-nas → type: sftp or smb

Step 10: Verify Everything Works

# 1. Claude Code responds:
claude --version

# 2. Memory is intact:
claude "What do you know about the Barney Ebsworth project?"

# 3. SSH to droplet works:
ssh root@143.198.182.180 "ls /var/www/barneyebsworth.com/sources/ | wc -l"

# 4. Sources are local:
dir C:\Users\USERNAME\websites\barneyebsworth.com\sources\

# 5. B2 sync works:
b2 ls walhus-private/barneyebsworth-sources/ | head -5

The Principles (carry to every machine)

1. Grep sources before writing any biographical claim. If it's not in a source file, say "I don't know."
2. Push everything to droplet + B2. Never leave files only on local disk.
3. Add blocked URLs to Paul's Jobs at barneyfrauenthal.com/paulsjobs/
4. Add every new person to the People database at barneyebsworth.com/people/
5. Be proactive. Build systems for recurring problems. Surface action items.
6. After compaction, read the recovery guide at .claude/projects/.../memory/barney-project-recovery.md FIRST.

Quick Copy Script (run on walhu to package everything)

@echo off
REM === Claude Config Backup Script ===
REM Run this on walhu to create a transferable package

set BACKUP=C:\Users\walhu\claude-transfer-package
mkdir "%BACKUP%" 2>nul
mkdir "%BACKUP%\.claude" 2>nul
mkdir "%BACKUP%\sources" 2>nul
mkdir "%BACKUP%\.ssh" 2>nul

REM Copy Claude config
xcopy "C:\Users\walhu\.claude\*" "%BACKUP%\.claude\" /E /I /Y

REM Copy SSH keys
copy "C:\Users\walhu\.ssh\id_*" "%BACKUP%\.ssh\"
copy "C:\Users\walhu\.ssh\known_hosts" "%BACKUP%\.ssh\"
copy "C:\Users\walhu\.ssh\config" "%BACKUP%\.ssh\" 2>nul

REM Copy source archive (large — 355MB)
xcopy "C:\Users\walhu\websites\barneyebsworth.com\sources\*" "%BACKUP%\sources\" /E /I /Y

REM Copy watcher scripts
xcopy "C:\Users\walhu\websites\.sync\*" "%BACKUP%\.sync\" /E /I /Y

REM Copy rclone config
copy "%APPDATA%\rclone\rclone.conf" "%BACKUP%\" 2>nul

echo.
echo === Package created at %BACKUP% ===
echo Transfer this folder to each new machine via USB or network share.
echo Then run the install steps above.
pause