Claude Build

1. Environment Overview

This is the WholeTech Network: 120 websites on a DigitalOcean droplet, a 14 TB Synology NAS holding 250,000+ original photos, two Beelink mini-PCs running automations, Backblaze B2 for cloud storage, and Claude Code orchestrating everything.

The 3-Tier Architecture

GOOGLE / EXTERNAL SOURCES
        ↓ (Takeout, Limitless API, scraping)

  HOT SPRINGS BEELINK (Windows 11)
  · Chrome for browser-auth downloads
  · Git Bash + Python 3 for scripts
  · Claude Code for orchestration
  · Persistent mapping to NAS as Z:\
        ↓

  SYNOLOGY DS1522+ NAS (14 TB)
  · Master archive for everything
  · SMB share as \\HS_DS1522plus\home
  · Mapped to Windows as Z:
        ↓

  BACKBLAZE B2 (cloud)
  · walhus-photos (public, web serving)
  · walhus-private (encrypted backup)
        ↓

  DIGITALOCEAN DROPLET (ubuntu-s-1vcpu-1gb-nyc1)
  · 120 websites live
  · Cloud photo grading (Claude API)
  · Password-protected private data viewers
  · Daily nightly backups at 3 AM

2. Hardware

Critical BIOS Settings (Hot Springs Beelink)

Must be set so the Beelink auto-restarts after power outages:

• Enter BIOS: Delete or F2 during boot
• Find "Restore on AC Power Loss" or "After Power Failure" → set to Power On
• Location: Advanced → Power Management
• Save with F10

Windows Auto-Login

1. Press Windows+R, type netplwiz
2. Uncheck "Users must enter a user name and password"
3. Apply, enter current password twice to confirm

Sleep Settings

Settings → System → Power & battery → Screen and sleep → "Never" for sleep (screen can turn off, system cannot).

NAS Auto-Restart

DSM → Control Panel → Hardware & Power → check "Restart automatically after a power failure"

3. Accounts and Credentials

All accounts required to rebuild the environment. Store these securely - never in a public repo.

Standard Credentials

B2 API Credentials

4. Domain Portfolio

120+ domains registered primarily through GoDaddy. All point to the DigitalOcean droplet (143.198.182.180) via A records.

Flagship Domains

• wholetech.com - Network hub, strategy pages
• walhus.com - Personal site, Paul's biography and portfolio
• convcast.com - Tech convention coverage (NEW April 13, 2026)
• austinspring.com - SpringNet heritage site, hosts /photos
• firth.com - Family history

Category Domains

DNS Configuration

Every domain needs these A records in GoDaddy DNS:

Critical: Delete any GoDaddy parking/forwarding records. Keep only the two A records above. SSL is handled by Let's Encrypt on the droplet.

5. Synology NAS Setup

The DS1522+ is the master archive for everything - photos, lifelogs, website backups, Google data.

Initial Setup

\\HS_DS1522plus\home\
├── photos\                   (final flattened photo archive)
├── google-photos\
│   ├── zips\                 (downloaded takeout zips, deleted after extraction)
│   ├── extracted\            (unpacked during processing)
│   └── old-attempt-apr11\    (historical)
├── google-takeout\
│   ├── zips\
│   └── extracted\
├── digitalocean\             (droplet backups)
│   ├── www\                  (120 website files)
│   ├── nginx\
│   ├── letsencrypt\
│   ├── backups\              (8 daily tarballs)
│   └── crontab.txt
├── lifelog\                  (Limitless pendant data, 57+ weeks)
│   ├── YYYY-MM-DD\           (one folder per week, Sunday-dated)
│   │   ├── audio\
│   │   ├── by-day\
│   │   ├── transcripts.json
│   │   └── _summary.txt
│   ├── exports\              (manual data exports)
│   └── fetch_limitless.py
├── beelink-archive\          (overflow from C: drive)
└── digitalocean\             (repeated - daily sync destination)

Mapping from Windows

net use Z: \\HS_DS1522plus\home /persistent:yes

Or through File Explorer: Map Network Drive → Z: → \\HS_DS1522plus\home → check "Reconnect at sign-in"

6. Beelink Setup

Windows 11 Pro box running Chrome, Git Bash, Python, and Claude Code. Two instances (Hot Springs primary, Cedar Creek backup).

Installation

net use Z: \\HS_DS1522plus\home /persistent:yes

pip install anthropic b2sdk

echo "sk-ant-api03-..." > ~/.anthropic_api_key
# For grading photos via Claude

7. DigitalOcean Droplet

Ubuntu droplet hosting all 120 websites, cloud photo grading, and private data viewers.

Software Stack

• nginx - web server for all 120 sites
• certbot - Let's Encrypt SSL certs, auto-renewal via cron
• Python 3 - cloud photo grading, parsers
• anthropic + b2sdk - via pip3 install --break-system-packages
• crontab - nightly backups, cert renewal

Nginx Configuration

Each website has a config in /etc/nginx/sites-available/DOMAIN symlinked to /etc/nginx/sites-enabled/.

Standard template:

server {
    server_name DOMAIN www.DOMAIN;
    root /var/www/DOMAIN;
    index index.html;
    location / { try_files $uri $uri/ =404; }
    listen 443 ssl;
    ssl_certificate /etc/letsencrypt/live/DOMAIN/fullchain.pem;
    ssl_certificate_key /etc/letsencrypt/live/DOMAIN/privkey.pem;
    include /etc/letsencrypt/options-ssl-nginx.conf;
    ssl_dhparam /etc/letsencrypt/ssl-dhparams.pem;
}

Private Pages (password-protected)

# Create htpasswd
htpasswd -bc /etc/nginx/.htpasswd-private paul testpass1944

# Add to nginx config
location /private/ {
    auth_basic "Private Area";
    auth_basic_user_file /etc/nginx/.htpasswd-private;
    try_files $uri $uri/ $uri/index.html =404;
}

SSL Setup (per domain)

certbot --nginx -d DOMAIN -d www.DOMAIN --non-interactive --agree-tos --email walhus@gmail.com

Nightly Backups (on droplet)

Crontab: 0 3 * * * cd /root/backups && tar -czf www-$(date +\%Y\%m\%d).tar.gz /var/www/ && ls -t www-*.tar.gz | tail -n +9 | xargs rm -f

Creates /root/backups/www-YYYYMMDD.tar.gz nightly, keeps last 8 days.

8. Backblaze B2 Storage

Account Setup

1. Sign up at backblaze.com, choose B2 Cloud Storage
2. Add credit card (first 10 GB free)
3. Create Application Key: App Keys → Add New Application Key → all buckets, read/write
4. Save keyID and applicationKey - shown ONCE

Buckets

Creating Buckets via SDK

from b2sdk.v2 import InMemoryAccountInfo, B2Api
info = InMemoryAccountInfo()
api = B2Api(info)
api.authorize_account("production", "KEY_ID", "APP_KEY")
bucket = api.create_bucket("walhus-photos", "allPublic")

Public URL Pattern

https://f000.backblazeb2.com/file/walhus-photos/{file_path}

9. Claude Code Installation

Claude Code is the AI orchestration layer. Runs on both Beelinks. Connects to Anthropic's Claude API and MCP connectors (Gmail, Calendar).

Installation

1. Download from claude.com/code or via npm install -g @anthropic-ai/claude-code
2. Sign in with Paul's Anthropic account
3. Claude Code uses the logged-in subscription for chat; separate API key for programmatic grading

Memory System

Persistent notes in C:\Users\walhu\.claude\projects\C--Users-walhu-websites\memory\:

• MEMORY.md - index of all memory files
• user_*.md - user preferences, role, etc.
• feedback_*.md - corrections Paul has given
• project_*.md - in-progress work, context
• reference_*.md - external system references

MCP Connectors

Connected at claude.ai/settings/connectors:

• Gmail (connector_uuid: f2146955-a798-40da-80f3-e41ac10482be)
• Google Calendar (connector_uuid: c279b02f-8049-4292-8c07-89415e696aa1)
• Slack (available but not actively used)

Scheduled Agents (Remote)

Managed at claude.ai/code/scheduled. Run on Anthropic's cloud, independent of local hardware.

• Takeout Email Monitor - checks Gmail every 4 hours for Takeout completion emails, creates Calendar event + Gmail draft when found

10. 120 Websites

All 120 sites live at /var/www/DOMAIN/ on the droplet. Standard template includes AdSense, Analytics, subnav bar, WholeTech branding.

Standard Site Features

Every site's index.html includes:

<meta name="google-adsense-account" content="ca-pub-7759195213529699">
<script async src="https://www.googletagmanager.com/gtag/js?id=G-MFQ0P2H8G8"></script>
<script>gtag('config','G-MFQ0P2H8G8');</script>
<script async src="https://pagead2.googlesyndication.com/pagead/js/adsbygoogle.js?client=ca-pub-7759195213529699"></script>

And the WholeTech lightning bolt "Return to network" badge:

<a href="https://wholetech.com" style="...">⚡ WholeTech</a>

Subnav Bar

Standard links: About, Resources, FAQ, News, Videos, Social, Sitemap

11. Photo Pipeline

The automated system that extracts photo zips, uploads to B2, grades with AI, and deploys galleries.

Scripts

Pipeline Flow

takeout.google.com
    ↓ Chrome downloads to Z:\google-photos\zips\
zip file complete (.zip, not .crdownload)
    ↓ auto-pipeline.sh detects it
unzip to Z:\google-photos\extracted\Takeout\Google Photos\
    ↓
upload-to-b2.py pushes new files to walhus-photos bucket
    ↓
grade-photos.py (Beelink) OR cloud-grade-photos.py (DO) grades each photo
    ↓
JSON manifest rebuilt with all grades
    ↓
Gallery HTML injected with manifest, deployed to austinspring.com/photos

12. Google Takeout Workflow

How to extract all Google account data - the one-time export that seeds the entire archive.

Parsers

13. Automated Systems

Windows Task Scheduler (Beelink)

Create with PowerShell:

$action = New-ScheduledTaskAction -Execute 'C:\Program Files\Git\bin\bash.exe' -Argument '-l -c /c/Users/walhu/websites/sync-droplet-to-nas.sh'
$trigger = New-ScheduledTaskTrigger -Daily -At 4:00AM
$settings = New-ScheduledTaskSettingsSet -StartWhenAvailable -DontStopIfGoingOnBatteries -AllowStartIfOnBatteries
Register-ScheduledTask -TaskName 'Sync Droplet to NAS' -Action $action -Trigger $trigger -Settings $settings -User $env:USERNAME -Force

Droplet Crontab

0 3 * * * cd /root/backups && tar -czf www-$(date +\%Y\%m\%d).tar.gz /var/www/ && ls -t www-*.tar.gz | tail -n +9 | xargs rm -f
0 2 * * * certbot renew --quiet

Limitless Lifelog Fetch

Script on NAS: Z:\lifelog\fetch_limitless.py

API key: ~/.limitless_api_key

Usage: python3 fetch_limitless.py (current week) or with date argument

Claude Code Scheduled Agent

Trigger: trig_01Ap71eVHgoVVMv7VxvUpPbc

Runs every 4 hours in Anthropic cloud. Checks Gmail for Takeout emails, creates Calendar alert + draft when found.

Manage: claude.ai/code/scheduled

14. All Scripts Reference

Every automation script, where it lives, and what it does.

15. Full Rebuild Playbook

Step-by-step to rebuild the entire environment from nothing. Assumes Paul has his existing Google, DigitalOcean, GoDaddy, and Backblaze accounts.

Phase 1: Foundation (Day 1)

net use Z: \\HS_DS1522plus\home /persistent:yes

Phase 2: Cloud Accounts

Phase 3: Restore Websites

apt update && apt install -y nginx certbot python3-certbot-nginx

scp -r /z/digitalocean/nginx/* root@143.198.182.180:/etc/nginx/

scp -r /z/digitalocean/letsencrypt/* root@143.198.182.180:/etc/letsencrypt/
# or re-issue:
certbot --nginx -d DOMAIN -d www.DOMAIN

Phase 4: Restore Data

cd /c/Users/walhu/websites
bash auto-pipeline.sh  # watches and processes zips as they land

Phase 5: Automation

16. Disaster Recovery

If the Beelink dies

• Websites keep running (they're on the droplet)
• Daily sync stops - no impact for a few days
• Replace Beelink, restore Claude Code memory from NAS backup, remap Z:

If the NAS dies

• Websites keep running (on droplet)
• Photos still accessible via B2 public bucket
• Restore from private B2 backup: lifelog/, scripts/, photo-grades.json
• Re-run Takeout to rebuild photo archive on new NAS

If the droplet dies

• All websites offline
• Create new droplet (same spec)
• SCP website files from NAS: /z/digitalocean/www/www/*
• SCP nginx config, letsencrypt certs
• Re-register A records if IP changed

If everything dies at once

• Google Photos still has the original photo library (5 TB cloud plan)
• B2 walhus-private has scripts, grades, lifelog transcripts
• B2 walhus-photos has safe-rated public photos
• GitHub has code (paulwalhus/wholetech-network)
• Follow full rebuild playbook (Section 15)