This is for co-ordinating instances of Claude or Codex, not something you do inside each instance.

Claude and Codex can also use the cook command to coordinate runs of other agents. This is similar to how you can describe a workflow to them of how to use subagents, and they'll try, but this gives them a reliable deterministic way to run those agents. An added benefit of having Claude/Codex/etc. use cook directly is that they are really good at analyzing the traces of what is happening inside of cook and after the fact.

Very nice. However, I do like to read every agent summary before letting them move on. I'm not sure I'd be able to apply this level of automation to many tasks.

I had to reread the first paragraph several times before I understood - the author was misuing a term.

> unexpected-behaviour.exe.dev

That is not a URL, that's a fully qualified domain name (FQDN), often referred to as just 'hostname'.

But that's one of the first things you fix in your CLAUDE.md: - "Only do what is asked." - "Understand when being asked for information versus being asked to execute a task."

This - per extensive experiments - works about as well as when I tell my wife to calm down

Asking might work better than telling

How do you do that???? Say the words but in the form of a question? I feel like that will go a lot worse than just telling (but nicely). I have a daughter too so I am genuinely willing to try anything

Please and thank you and make sure you’re addressing the behaviour and not the person.

But is anthropic trying to solve it? The current permissions solution is unbelievably poor for a product with this much traction.

They are releasing auto-mode soon. But that won't improve the underlying permission system, rather, it'll just delegate decisions to Claude. That's better than --dangerously-skip-permissions, but not great for those that want granular controls and are sensitive to the extra tokens spent.

Sensitive lad.

You can also turn off tips, but that's quite handy to learn when a new feature pops up

Oh imagine LLMs insisting on tips! 20% “voluntary”.

The entire kernel on every arch is 40 million lines, but the kernel running on your desktop is probably less than 2 million of those lines.

it's more the way they do them.. you've used them right?

Sure but I don't find them irreplaceable. Actually anthropic models have dropped out of my top ten usage this month. I only use opus occasionally for writing plans, its been pretty unreliable at executing.

The US is such a shit show. Personally I hope this doesn't affect Anthropic's growth and development because I quite enjoy using their products and see them evolve.

I find it very intuitive as is