Quick start

This takes about two minutes. You’ll need an API key from one AI provider. If you don’t have one, these have free options:

Provider	Free option	Get a key
Groq	Free tier, very fast	console.groq.com/keys
Google Gemini	Free tier	aistudio.google.com/apikey
Ollama	Free, runs on your computer, no key	Ollama guide

Your first chat

1. Open kunji.

   kunji

   The first time, kunji helps you add a provider.

2. Pick a provider with the arrow keys and press Enter.

3. Paste your API key. It’s hidden as you paste. kunji checks it with the provider, then saves it in your operating system’s keychain, never in a text file.

4. Choose a default model. Start typing to search the list, then press Enter.

5. Start a chat. Choose Start a new chat in the menu, type a question and press Enter. The answer appears as it’s written, with its cost underneath:

   › explain recursion in one line
     A function that solves a problem by calling itself on a smaller piece of it.
     llama-3.3-70b-versatile · 18 in · 21 out · <$0.0001 · 0.4s

6. Leave the chat with Ctrl+D or /exit. You’re back in the menu. Choose Exit to close kunji.

Prefer typing commands?

The same steps without the menu:

kunji keys add groq     # add a key
kunji chat              # start chatting

Things to try next

• Type / in a chat to see every command.
• Switch model mid-chat: /model groq:llama-3.3-70b-versatile (Tab completes names).
• Write a commit message: kunji @commit inside a git project.
• Continue tomorrow where you left off: kunji chat -c.
• See what you’ve spent: kunji usage.

Learn more

• The menu
• Chatting
• Command reference