So, in my store (https://actions.work/store/), next to my apps, I now also sell "Tokens of Appreciation" for €2 a pop. I described them as such:

> Technically these tokens do nothing. Emotionally they mean the world to me.

Let's see how it goes. I don't expect anything, which is probably a […]

@ctietze I usually use Claude Desktop for research, which does all of that pretty well. When I'm done w/ researching, I have it create a report which I can then hand over to Claude Code. I might be holding it wrong in that regard, too, but it's what works for me.

@ctietze Hmm. Just thinking out loud here: kagi_search_fetch only returns short JSON snippets, so I doubt it's those that fill up context. But if your LLM then fetches and examines those single results, it has to work with them somewhere, so context is used. You could have it create a "web […]

PS: I liked it.

@denis Das wäre dann Corned Beef ("Kohrnettböff", wie mein Vater es nennt).

Keine Ahnung, probiere es aus. Bitte livestreamen, damit kein Wissen verloren geht, falls Du es nicht überleben solltest. 🖖🏼

@denis ❤️ ist darin. Herz. Also Innereien.

@ctietze I have no idea, I just wanted to be part of the conversation

@ctietze YES