← All products
🚚

AI Agent Migration Pack

Your AI agent runs on your laptop. It dies when you close the lid. This kit moves it to dedicated hardware that runs 24/7, with discovery scripts, path rewriting, rollback planning, local-model resource guardrails, and cross-platform setup paths for Mac Mini, Linux server, or cloud VM.

✓From a real 72-hour production migration, now updated with Mac Mini resource guardrails.
Updated May 11, 2026 · Added local-model sizing, thermal/resource guardrails, and staged rollback checklist from the Mac Mini lessons.

Paid Digital Thoughts subscriber?

Yearly subscribers: all products free ($1150.94+ value)

Monthly subscribers: 1 free product per month — subscribe from $29.99/mo

Claim your free copy →
$49.99One-time purchase. Lifetime updates.

What changed in this update

  • ✓Added local-model-resource-guardrails.md
  • ✓Added rollback-cutover-plan.md
  • ✓Corrected path-rewriter and health-monitor references in the docs

Best for

  • +Moving an agent from a daily laptop to a Mac Mini, Linux box, or cloud VM
  • +Agents with cron jobs, LaunchAgents, state files, credentials, and hardcoded paths

Not for

  • -One-off scripts that do not need 24/7 uptime
  • -Hardware moves without SSH/admin access to the target machine

What you get

+Discovery-first approach: agent scans your current setup and builds a custom migration plan
+Cross-platform: Mac Mini, Linux server (Ubuntu/Debian), cloud VM, Windows WSL
+Memory and state migration: move sessions, context, preferences, and learned patterns intact
+Automation migration: LaunchAgents, cron jobs, systemd units with automatic path rewriting
+Headless operation: virtual displays, screencapture, UI automation without a monitor
+Local-model resource guardrails: CPU, memory, swap, disk, and thermal preflights before unattended runs
+Rollback cutover plan: mirror, shadow, primary, and roll back if the first 24 hours fail
+Security and permissions: TCC databases, SSH hardening, credential vaults, firewall rules
+Health monitoring: watchdog scripts, heartbeat checks, recovery procedures, cost tracking
+Agent-ready format: feed the guide to your AI agent and it walks you through each step

Package includes

  • •Comprehensive Markdown guide (~40 pages, 12 chapters): architecture, planning, execution, verification
  • •discover.sh: scans current machine, inventories agents/crons/paths/dependencies (outputs JSON)
  • •migrate.sh: orchestrates the full migration with dry-run mode
  • •verify.sh: post-migration health check (services running, paths valid, permissions correct)
  • •scripts/path-rewriter.sh: finds and rewrites every hardcoded path with backup
  • •Platform-specific setup scripts (macOS, Ubuntu/Debian, cloud VM)
  • •local-model-resource-guardrails.md and rollback-cutover-plan.md
  • •Headless operation guide (BetterDisplay, xvfb, VNC, virtual displays)
  • •LaunchAgent/cron/systemd migration templates with automatic conversion
  • •Security hardening checklist (SSH, TCC, firewall, credential management)
  • •Health monitor template (Python watchdog with alerting)
  • •Cost calculator (hardware vs cloud ROI comparison)
  • •QUICKSTART.md: working migration in 30 minutes (simple case)
  • •CHANGELOG.md

FAQ

Does this only work with Claude Code?

No. The migration scripts work with any AI agent system: Claude Code, OpenClaw, Aider, Continue, or custom setups. The discovery script finds whatever is running on your machine.

Do I need a Mac Mini specifically?

No. The kit covers Mac Mini, any Linux server (Ubuntu/Debian), cloud VMs (DigitalOcean, AWS, Hetzner), and Windows WSL. Each platform has its own setup path.

How long does migration take?

Simple setups (single agent, few automations): 30-60 minutes with the QUICKSTART. Complex setups (multiple agents, many cron jobs, headless requirements): 2-4 hours with the full guide. The discovery script tells you which path you need.

What if my agent has hardcoded paths?

The discover.sh script specifically catches hardcoded paths and generates a replacement map. This was one of the biggest silent failure modes in the real migration this kit is based on.

Can I feed this to my AI agent?

Yes. The guide is structured for AI agents to follow step-by-step. The discover.sh output is machine-readable JSON. You can paste the agent-guide into your agent and say "migrate me."

Related reading on Digital Thoughts:

Secure checkout by Stripe. Instant download + guided Claude Code setup.