OpenClaw on Mac Mini

Video Guide: Full OpenClaw Installation & Configuration

Watch the complete walkthrough showing physical setup, wiring, initial macOS configuration, OpenClaw install, API keys, Telegram bot connection, security notes, and more.

Video: Setting up a shared Mac Mini for OpenClaw (formerly Clawdbot), including hardware wiring, naming convention ("Trina Baldwin"), Ethernet preference, tool installs, Venice AI / Kimi model, Telegram BotFather integration, gateway service, terminal commands, hidden files view, and important security warnings.

Overview: Setting Up a Shared Mac Mini Server for OpenClaw

Physical & Hardware Preparation

• Use a low-cost Mac Mini as a dedicated, always-on machine for the team/community (named e.g. "Trina Baldwin" with a sticker label).
• Position it under the desk or in a clean spot; repurpose an existing monitor, HP keyboard + mouse (with USB adapter if needed).
• Connect via long Ethernet cable directly to router — avoids Wi-Fi instability, IP/MAC changes, and disconnections.
• Manage cable routing neatly to minimize clutter; swap HDMI from another machine if necessary.
• Power on and begin standard macOS setup in English.

Initial macOS & Account Configuration

• Skip or minimize tracking/privacy prompts where possible; enable location if needed for full functionality.
• Immediately install Google Chrome (preferred for AI tooling compatibility) over Safari.
• Sign into a dedicated Google account (treat as team/company account — e.g. shared worker credentials).
• Use secure methods to transfer long passwords/API keys (e.g. internal platform like webchat.gfavip.com / GFA instead of manual typing or risky channels).
• Verify email, iTunes/App Store login, and complete basic setup.

OpenClaw Installation & Basic Setup

• Install via the official one-liner in Terminal (from openclaw.ai): curl -fsSL https://openclaw.ai/install.sh | bash
• Handle permission prompts during install (may require admin escalation via Terminal commands like granting sudo rights).
• Add LLM API key (e.g. Venice AI, Anthropic/Claude, OpenAI, Kimi — set Kimi 2.5 as default in this example).
• Create Telegram bot via @BotFather: /newbot → name ending in "bot" → copy API token → paste into OpenClaw config.
• Install optional skills/tools (e.g. GOG for Google email integration via NPM — follow prompts).
• Enable gateway service (web UI + communication layer); choose web chat interface for browser-based interaction.
• Test initial chat in the gateway UI (agent responds via selected model).

Terminal & File System Tips

• Use Terminal for management: check status, logs, config, restart daemon (commands may vary: openclaw status, clawdbot status, etc. — project naming has evolved).
• Reveal hidden files in Finder: defaults write com.apple.finder AppleShowAllFiles YES && killall Finder
• Access OpenClaw folders: ~/agents, ~/canvas, ~/cron, ~/identity, ~/logs, ~/devices, ~/.openclaw (secrets, models, main agent config).

Critical Security Notes

• Do NOT expose the gateway publicly over LAN/Wi-Fi/Internet — default setup may bind to local IPs; lock it down immediately.
• Run community-recommended security checks (search Twitter/X or blogs for latest OpenClaw hardening guides).
• If someone reaches the IP/port, they could abuse the agent — prioritize firewall rules, auth, no public discovery.
• Next steps after basic install: finish Telegram connection, build agent identity/soul (markdown files), train it, add more tools, monitor logs.

OpenClaw turns a simple Mac Mini into a powerful, private 24/7 AI teammate. Start small, secure first, then expand integrations.