The Path › Chapter 3 of 9
Talk to it
one question, then a conversation
Teach
A prompt is a request, not a command
When you type to a brain you are not giving an instruction a computer obeys; you are handing over a request it interprets. Two things follow. First, how you ask changes what comes back — a vague question gets a vague answer, and "explain it like I have never seen a terminal" genuinely works. Second, the brain knows only what it was given: the model file and your message. It cannot see your files, your calendar, or the time unless something hands those to it. Chapter 4 is about handing things to it — that is what tools are.
Two ways to talk
One question. adk backend test sends a single fixed prompt to whatever brain the kit is pointed at and prints the reply with the provider, the model and a token count. It is the smoke detector: if this says Status: OK, everything above it in the stack is fine.
A conversation. adk start opens a chat. Run it inside a folder and it also reads that folder — Python files get indexed — so you can ask "what does this project do?" and get an answer grounded in what is actually there. The banner at the top tells you which brain it found. Type /quit to leave.
Local or cloud is per task, not forever
Everything so far runs on your machine with no account. Aitherium also hosts bigger brains, and adk login connects you to them. It is optional in this guide, and it stays optional: every later chapter works offline.
If you do log in, notice what does not happen: you never type a password into the terminal. The kit prints a short code and a web address; you open the page, type the code, and come back. That is a device code login — the same thing a TV does when you sign in to a streaming app — and it means the terminal never holds your password at all.
adk whoami tells you which state you are in. Both answers are fine.
Do
adk backend test
You should see: Provider: ..., a one-line Response: from your brain, and Status: OK
If not: If it says FAILED, the brain from chapter 2 is not running. adk quickstart-local again restarts it; adk backend status shows what the kit expects to find.
adk start is a chat with your brain that also knows the folder you ran it in. Python files get indexed, so 'what does this project do?' gets a real answer.
adk start
You should see: a small banner - Workspace, Directory, LLM: your local brain and its address - then a You > prompt. Type a question; type /quit to leave.
If not: If the LLM line names the cloud or says nothing was found, the brain is down - see the step above. Run adk start inside any folder; it reads that folder so it can answer questions about it.
adk login
You should see: a short code and a URL; open the URL, type the code, come back - Logged in as ...
If not: Logging in is optional. It connects you to hosted brains and your workspace. Stay local if you prefer - everything in the next chapters works offline too.
adk whoami
You should see: your account, or Not logged in
If not: Both answers are fine. This just tells you which you are.
Check you are done
adk backend test
You should see: Status: OK
What you learned
- A prompt is a request; the brain's answer depends on what it was given, which is why chapter 4 gives it tools.
- Local vs cloud is a choice you make per task, not once:
adk backend statusshows which is current. - Your login is a device code you type into a web page, never a password typed into a terminal.