Local Mac Mini Setup Guide

Why a Mac Mini

An always-on agent needs a machine that is always on. I tried an Amazon EC2 instance first: latency, SSH complexity, dropped sessions, and a monthly bill that kept climbing. Moving to a Mac Mini on my desk fixed all four.

The Mac Mini has become the default hardware for running personal agents. Its compact size, Apple Silicon efficiency (3–7 watts at idle), and unified memory make it a quiet, always-on server at a fraction of cloud hosting costs. This guide walks you from unboxing to a hardened machine that is ready for an agent.

Part 1: Hardware & Prerequisites

Recommended Hardware

The base M4 Mac Mini at $599 is enough if your agents call cloud models. Buy more memory only if you plan to run local models (see Part 7). At roughly $1–2/month in electricity, it pays for itself within months versus equivalent cloud hosting.

What You'll Need Before Starting

• An Anthropic API key or subscription (recommended), or access to another frontier model provider
• A messaging channel for talking to your agent, such as a Telegram bot token from @BotFather
• A Tailscale account (free tier) for secure remote access
• Approximately 2 hours of dedicated setup time

Part 2: Initial macOS Configuration

Fresh Install or Factory Reset

If you're repurposing an existing Mac Mini, perform a full factory reset: System Settings → General → Transfer or Reset → Erase All Content and Settings. Starting clean eliminates leftover software and credentials that an agent could inadvertently access.

Create Two Accounts: Separation of Privilege

Before installing any agent software, create a separation of privilege on the Mac Mini itself. This is one of the most effective containment strategies you can implement, and most guides skip it entirely.

Admin Account (for installs only): Used exclusively for installing software (Homebrew packages, runtimes, macOS updates). It does not run your agents.

Standard (Non-Admin) Account (for running agents): This is where your agents live and run. A standard account cannot install system software, modify system files, or escalate privileges. If an agent gets tricked by a prompt injection or a malicious plugin, the blast radius is contained.

macOS Setup Wizard Decisions

Configure for 24/7 Operation

Prevent the Mac Mini from sleeping and ensure it auto-restarts after power failures:

sudo pmset -a sleep 0 disksleep 0 displaysleep 0
sudo pmset -a hibernatemode 0 powernap 0
sudo pmset -a standby 0 autopoweroff 0
sudo pmset -a autorestart 1

Verify with: pmset -g — all sleep-related values should be 0. Enable wake for network access so Tailscale stays connected: System Settings → Battery → Options → Wake for network access: ON.

Enable the macOS Firewall

The built-in firewall is off by default. Turn it on:

1. System Settings → Network → Firewall → Toggle ON
2. Click Options: Block all incoming connections: ON
3. If you later need Tailscale or SSH, add exceptions specifically for those apps

Disable SSH Password Authentication

If Remote Login (SSH) is enabled, disable password-based authentication and allow only key-based auth:

sudo nano /etc/ssh/sshd_config
# Set these values:
PasswordAuthentication no
ChallengeResponseAuthentication no
UsePAM no

# Restart SSH:
sudo launchctl unload /System/Library/LaunchDaemons/ssh.plist
sudo launchctl load /System/Library/LaunchDaemons/ssh.plist

Part 3: Install Prerequisites

Install Homebrew

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Follow the on-screen instructions to add Homebrew to your PATH.

Install Your Runtime's Dependencies

Most agent runtimes need Node.js, Python, or both. Install whatever yours requires from the admin account, on current LTS versions, and keep them patched. Security fixes land in the runtime as often as in the agent software.

brew install node python git
node --version
python3 --version

Install Tailscale

Tailscale provides secure, zero-configuration VPN access to your Mac Mini from anywhere:

brew install tailscale
# Also install the Tailscale app from the Mac App Store for GUI management

Plug In the HDMI Dummy Plug

On recent macOS versions, headless mode causes issues with Screen Recording permissions and screen capture functionality. Plug an HDMI dummy dongle ($8–10) into one of the HDMI ports before proceeding. This tricks macOS into thinking a display is connected.

Part 4: Principles for the Agent Install

Install the agent software itself by following its own docs. Whatever you run, these are the decisions that matter:

Run it as the standard user. Install with the admin account if you must, but the agent process runs under the non-admin account you created in Part 2.

Use a frontier model. Stronger models resist prompt injection better. Weaker or older models are more easily manipulated, and when your agent has shell access, that matters.

Bind to localhost only (critical). If your runtime exposes a local server or dashboard, it must only be reachable from the Mac Mini itself, never from the network. Loopback, not 0.0.0.0. Remote access comes through Tailscale in Part 5, not an open port.

Lock the messaging channel to you. Allowlist your own user ID on Telegram (or whatever channel you use) so strangers can't talk to your agent. Unknown senders should be ignored or require explicit approval.

Start it on boot. Run the agent as a LaunchAgent (or whatever service mode your runtime provides) so it starts on boot and restarts on crash.

Set the Agent's Identity

The first instructions you give an agent shape its behavior going forward. Be deliberate:

Your name is [Agent Name]. I'm [Your Name].
Be direct, concise, and honest.
If I ask you to do something risky, push back and tell me why.
Don't sugarcoat or over-explain.
Core rules:
- Never delete files, send messages, or run commands with side
  effects without confirming with me first.
- Ask before acting on anything destructive or irreversible.
- Keep responses short unless I ask for detail.
- Flag security concerns proactively.

Lock Down Config and Credential Files

Agent config directories hold API keys, bot tokens, and session data. Restrict them to the account that runs the agent:

chmod 700 ~/path/to/agent-config
find ~/path/to/agent-config -type f -exec chmod 600 {} \;

Set API Spending Limits

An out-of-control agent or a prompt injection attack could rack up enormous API bills. Set spending limits directly with your model provider as a safety net. Anthropic: set monthly usage limits in the Anthropic Console under Billing. OpenAI: set hard monthly caps in the OpenAI dashboard.

API Key Management

Never scatter API keys across .env files or shell history. Two recommended approaches:

Option A: Bitwarden CLI (Recommended). Install the Bitwarden CLI, log in, and unlock your vault. Your agents pull secrets on demand.

Option B: macOS Keychain. Store API keys in the local Keychain (encrypted via the Secure Enclave). Credentials never leave the device if iCloud Keychain is disabled.

Network Security Checklist

• ✓Agent server or dashboard bound to localhost only — never 0.0.0.0
• ✓macOS firewall is ON with incoming connections blocked
• ✓No agent ports forwarded to the internet via your router
• ✓Remote access is exclusively via Tailscale or SSH tunnel
• ✓Your home router's admin password has been changed from the default
• ✓UPnP is disabled on your router

Part 5: Secure Remote Access with Tailscale

You'll want to reach your Mac Mini and any agent dashboard from your phone or laptop. Never expose a port directly to the internet. Use Tailscale instead.

Set Up Tailscale

1. Install Tailscale (done in Part 3) and sign in
2. On your Mac Mini, authenticate: tailscale up
3. Install Tailscale on your phone/laptop and sign in with the same account
4. Verify connectivity: tailscale status

Configure Tailscale Serve

If your agent has a local dashboard, publish it to your tailnet only. Replace PORT with the port your runtime uses:

tailscale serve http://127.0.0.1:PORT
# Access from any device on your tailnet:
# https://[mac-mini-hostname].tail[xxxxx].ts.net

Alternative: SSH Tunnel

ssh -N -L PORT:127.0.0.1:PORT user@your-mac-mini.local
# Then open http://localhost:PORT

Part 6: Plugin & Skill Security

Skills, plugins, and MCP servers extend what your agent can do. They also run with your agent's access. Community marketplaces have shipped malicious packages designed to steal credentials, and vetting is thin.

Golden Rules for Skills

1. Read the source code before installing anything. Treat community skills with the same skepticism as random npm packages from an unknown author.
2. Start with zero third-party skills. Use only the built-in tools initially. Add third-party skills one at a time after thorough review.
3. Check the author and track record. Popularity alone doesn't guarantee safety, but brand-new packages from new accounts are especially risky.
4. Look for credential access. If a skill requests API keys or file system access disproportionate to its stated purpose, do not install it.
5. Sandbox anything you don't fully trust.

Part 7: Ongoing Maintenance & Monitoring

Update Schedule

Backups

Time Machine will back up the whole machine, including your agents' config and memory. Additionally, keep periodic exports of your config on an encrypted external drive, back up API keys separately in your password manager, and document your skill setup so you can rebuild quickly if needed.

What to Do If Compromised

1. Stop the agent process immediately
2. Disconnect from the network (Wi-Fi and Ethernet)
3. Review the agent's logs for suspicious activity
4. Rotate ALL credentials: API keys, bot tokens, and any accounts the agent accessed
5. Check your messaging channel for senders you didn't approve
6. Rebuild in an isolated environment with proper security controls before restarting

Part 8: Optional Enhancements

Local AI Models with Ollama

Run AI models entirely on your Mac Mini with no cloud API costs. Apple Silicon's Metal GPU acceleration makes local inference surprisingly fast.

brew install ollama
ollama pull llama3.1

Container Isolation

For an extra layer of isolation, run your agent inside a container using Docker Desktop or OrbStack. Run it as a non-root user and mount only the directories the agent actually needs.

Scheduled Routines

The biggest unlock of an always-on machine is that agents can work on a schedule: a morning brief, an inbox sweep, a weekly memory rollup. Use your runtime's scheduler, or plain cron or launchd, to run them in isolated sessions.

Security Checklist

• ✓Dedicated non-admin macOS user for running agents
• ✓FileVault encryption: ON
• ✓macOS firewall: ON
• ✓Agent server bound to localhost only
• ✓Messaging channel allowlisted to your IDs only
• ✓Frontier model (Claude or equivalent)
• ✓API spending limits set with provider
• ✓No agent ports exposed to the internet
• ✓Remote access: Tailscale Serve or SSH tunnel only
• ✓Agent software and runtimes on current versions
• ✓File permissions: 700 on config directories, 600 on config files
• ✓Third-party skills: Source code reviewed before installation
• ✓API keys: Stored in password manager, not .env files
• ✓Router UPnP: Disabled

Final Word

An always-on agent is genuinely useful. It's also genuinely risky if you're careless. Prompt injection, malicious plugins, and exposed instances are all real, and all have happened to people who skipped the boring parts of setup.

Every configuration decision in this guide was filtered through one question: what's the worst that could happen? Set it up right the first time. Start locked down. Open things up only when you understand exactly what you're exposing.