I just got this working. The workflow is a little wonky to me, but better than I've seen from other AI-in-the-terminal solutions:

Setup:

1. Go to Settings (⌘+,)

2. General > AI. Put in your OpenAI key. I went with GPT-4o. Make sure you set the token limit to 4,096 or it wont go.

Using:

1. At the terminal window, hit Edit > Engage Artificial Intelligence, or just hit ⌘+Y.

2. Type in the command you want, then hit OK or Shift+Enter.

3. On the result window you can hit Shift+Enter again to paste the result into your terminal.

The whole Shift+Enter workflow is odd to me. I'd prefer just Enter. However, I like that I know exactly what's being sent to OpenAI. The integration doesn't try too hard to send everything over.

It's good to see iTerm get some updates too!

More discussion is over here: https://news.ycombinator.com/item?id=40425428

I'd love to see the ability to hook into my self-hosted ollama (llama3) server running in my homelab. I'm not thrilled about giving OpenAI all my terminal history.