# Quickstart

Run your first CrabTalk agent in under a minute — start the daemon and chat.

This guide gets you from zero to a working agent in under a minute.

## 1. Start chatting

Just run `crabtalk`:

```
crabtalk
```

On first run, it scaffolds the config directory at `~/.crabtalk/`, prompts you to configure a model, starts the daemon in the background, and drops you into an interactive REPL. Type a message and press Enter to chat.

## 2. Manage the daemon

The daemon runs as a background service. You can manage it explicitly:

```
crabtalk --start       # start daemon without entering chat
crabtalk --stop        # stop daemon
crabtalk --restart     # restart daemon
crabtalk --foreground  # run in foreground (for debugging)
crabtalk --reload      # hot-reload config
```

## 3. Configure models

To add or edit model providers later:

```
crabtalk config
```

See [providers](/content/docs/crabtalk/providers/index.html) for all supported models and API standards.

## What's next

- [Set Up Telegram](/content/blog/setup-telegram/index.html) — connect a Telegram bot in 5 minutes
- [Set Up Search](/content/blog/setup-search/index.html) — give your agent web search
- [Configuration](/content/docs/crabtalk/config/index.html) — customize directories, models, and services

### On this page

1. Start chatting  
2. Manage the daemon  
3. Configure models  
What's next