Smart people online are talking about loops. Frankly I don’t understand what these things are in terms of tangibles.
I use opencode all the time. Coworkers using codex, some claude.
But what is a loop? If you do use loops could you share explicitly in tangibles what it is and how you use it? Even a simple example would be tremendous I’m so lost haha
Just came upon a short that explains it and shows a simple use:
also, Geoffrey Huntley started the trend I believe:
Haven’t actually tried it yet so can’t say much more. I do have plans, it involves utilizing my rtx 4090. I’ll make some overnight experiments, see what happens. Generally, I do like being in the loop so… that’s why I haven’t tried it yet. 
In terms of tangibles, worth reading:
https://x.com/addyosmani/status/2064127981161959567?s=46
I personally haven’t felt much of a need to play around with this kind of setup yet, but I have that sense of inevitability that we’ll probably all be working this way in a years time, as the models, tooling etc all builds up around us.
You have to use map in a functional language.
bugs = [...]
agent = Enum.random(~w[claude codex])
Enum.map(fn bug -> System.cmd(agent, ["fix #{bug}"] end)
So a agent on a cronjob?! Sounds like a good way to generate steady token usage. Feels like there could be smarter systems doing the scheduling cheaper, while only talking to the LLM when necessary.
Somebody think of the shovel sellers!
© Copyright Elixir Forum | Terms | Privacy & Cookies
