Your agent understands two languages. The first one is the usual: you talk to it normally, like to a person —"summarize this email for me", "find me a hotel in Cadiz"— and it answers.
The second one is commands: quick, direct orders that start with a slash /. Instead of explaining what you want, you press a "button".
Look at the difference:
- If you type "which model are you using?", it replies with a sentence.
- If you type
/status, it hands you the full tech sheet in one go.
Commands are your agent's way of getting straight to the point. And no, it doesn't take offense when you give it orders: it actually loves them, that's its thing.
The golden rule: one command, one message
Send each command on its own, in its own message. Don't tuck it into the middle of a sentence.
✅ Good:
/status🚫 Better not:
hola, mira /status y dime algoThat way you keep it from getting muddled between "what you're telling it" and "what you're ordering it to do".
Telegram and the web: the same agent, two controllers
You can talk to your agent from Telegram or from its web panel (what we call the UI or webchat). Commands work in both places when you type them out as text.
The difference: on Telegram you type them; on the web panel there may also be buttons and menus that do the same thing without typing. For example, where on Telegram you'd type /model, on the web you might have a model selector.
Your hablo agent comes with its web panel and its Telegram already set up. You don't have to install or connect anything weird: you log in and use it. We take care of making the "controller" work.
A cheat sheet for the rest of the chapter
In the upcoming lessons you'll see each command with a color label, so you know at a glance how much "weight" it carries:
- 🟢 Safe: just looks up information. Breaks nothing.
- 🟡 Changes the conversation: modifies how your agent works in this chat.
- 🔴 Powerful / advanced: use it wisely (or leave it to support).
Let's dive into the ones you'll use every day. 👇