Step 1 · package manager / runtime
Get Node.js on the machine
| Platform | Command |
|---|
| Windows | winget install OpenJS.NodeJS.LTS (Windows Terminal + winget built in) |
| Mac | /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)" then brew install node |
| Linux / droplet | curl -fsSL https://deb.nodesource.com/setup_lts.x | sudo -E bash - then sudo apt install -y nodejs |
| Android (Termux from F-Droid) | pkg install nodejs git |
| Synology / Docker | sudo docker run -it node:lts bash |
The coding agents
Install, run, check
| Tool | Install | Run | Check |
|---|
| Claude Code | npm install -g @anthropic-ai/claude-code | claude | claude --version |
| Codex | npm install -g @openai/codex | codex | codex --version |
| Gemini CLI | npm install -g @google/gemini-cli | gemini | gemini --version |
| Hermes Agent ✦ | curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash (Windows: run inside WSL2) | hermes --tui | hermes --version |
💡Note: Claude Code, Codex, and Gemini install the same way anywhere Node is present (Windows, Mac, Linux, Termux, a container). Hermes Agent (Nous Research) uses its own installer — native on Mac/Linux, inside WSL2 on Windows — and is model-agnostic (works with Nous, OpenAI, Anthropic, OpenRouter).
Tailscale (private network)
One login on every device
| Platform | How |
|---|
| Windows | winget install tailscale.tailscale (or tailscale.com/download/windows) |
| Mac | brew install --cask tailscale (or Mac App Store) |
| Linux / droplet | curl -fsSL https://tailscale.com/install.sh | sh then sudo tailscale up (add --ssh on servers) |
| Android / iOS | official "Tailscale" app from Play Store / App Store |
| Synology | Package Center → install "Tailscale" |
| Check | tailscale status |
Ollama (local models)
Run AI on your own hardware
| Platform | Install |
|---|
| Windows | winget install Ollama.Ollama (or ollama.com/download/windows) |
| Mac | brew install ollama (or ollama.com/download/mac) |
| Linux / droplet | curl -fsSL https://ollama.com/install.sh | sh |
| Synology / Docker | sudo docker run -d --name ollama -p 11434:11434 -v /volume1/docker/ollama:/root/.ollama ollama/ollama |
| First model (everywhere) | ollama run llama3.2 (type /bye to exit) |
| Share over your network | set OLLAMA_HOST=0.0.0.0:11434, restart Ollama; reach it at that machine's Tailscale address, port 11434 |
The tool landscape at a glance
What's out there
The short version — see the full map and how they interrelate at /landscape/.
| Category | Tools |
|---|
| Coding agents (terminal) | Claude Code, Codex, Gemini CLI, Hermes Agent, Aider, Cline |
| Coding in an editor | Cursor, Windsurf, GitHub Copilot, Continue |
| Chat assistants | Claude, ChatGPT, Gemini, Perplexity |
| Run models locally | Ollama, LM Studio |
| Private network glue | Tailscale |
| Connect tools to your data | MCP (Model Context Protocol) |
🔑Login: the three coding agents open your browser to sign in the first time (Anthropic, OpenAI, Google accounts). On a server they print a URL to open on your laptop. Ollama needs no account.