Quick Setup

Three steps to a working AI assistant.

Step 1: Run the Setup Wizard

astonish setup

This interactive wizard walks you through selecting an AI provider and entering your API key. Supported providers include:

• OpenAI
• Anthropic
• Google Gemini
• Groq
• Ollama (local, no API key needed)
• OpenRouter
• xAI
• DeepSeek
• And more

Configuration is stored in ~/.config/astonish/. You can re-run astonish setup at any time to change providers or update keys.

Step 2: Install and Start the Daemon

astonish daemon install && astonish daemon start

This installs Astonish as a background service (launchd on macOS, systemd on Linux) and starts it immediately. The daemon is the foundation of Astonish — it powers everything:

• Serves the Studio web UI at http://localhost:9393 — a full visual interface for chat, settings, flow design, and fleet management
• Listens on communication channels (Telegram, Email) once configured
• Runs scheduled tasks on a cron schedule
• Manages fleet sessions for multi-agent teams

Open http://localhost:9393 in your browser to access Studio. On first visit, you will be prompted to create a password.

Check status at any time:

astonish daemon status

Step 3: Start Chatting

Now that the daemon is running, you can chat through Studio at http://localhost:9393, or use the CLI:

astonish chat

Try a few things to see what it can do:

• “Read the README in this directory and summarize it.”
• “Search the web for the latest Go release.”
• “List all files larger than 10MB in my home directory.”
• “Store my GitHub token as a credential.”

The agent has access to 74+ tools out of the box — file operations, web search, browser automation, email, credential management, and more. It will ask for confirmation before running anything destructive.

Next Steps

• Choose your interface — CLI, Studio, Telegram, or all of them
• Connect Telegram for mobile access
• Explore built-in tools — 74+ tools the agent can use