Installation

Prerequisites

• Node.js 22 or later
• An API key from OpenAI or Anthropic — or a locally running Ollama instance for fully private, offline inference
• A Telegram bot token (optional but recommended for first setup)

Install

npm i -g airupt

Setup Wizard

Run the interactive setup wizard:

airupt setup

The wizard will ask for:

1. LLM provider — Claude (Anthropic), OpenAI, or Ollama (local)
2. API key — stored locally in ~/.airupt/config.yml, never transmitted (not required for Ollama)
3. Channel — Telegram bot token, WhatsApp, Discord, or skip for now
4. Assistant name — defaults to "Airupt"

Setup takes under 2 minutes. When done, your config lives at ~/.airupt/config.yml.

Running Fully Local with Ollama

Airupt supports Ollama as an LLM provider, letting you run the assistant entirely on your own hardware with no API keys and no data leaving your machine.

1. Install Ollama and pull a model:

   ollama pull llama3.2

2. Run airupt setup and choose ollama as your primary provider.
3. Confirm the base URL (http://localhost:11434) and model name (llama3.2).

Ollama usage costs $0 — the cost tracking in airupt status will always show 0 for local inference.

Start the Assistant

airupt start

Airupt runs in the foreground. Use airupt start --daemon to run as a background process.

First Chat

Open Telegram (or whichever channel you configured) and send a message:

Hello! What can you do?

Airupt will respond with a summary of available skills and how to use them.

Verify Everything Works

airupt status

Shows which channels are connected, active skills, and current cost totals.

Next Steps

• Full configuration reference
• Set up additional channels
• Explore built-in skills