Making the behavior for "I disagree" and "this is erroneous" the same seems like a problematic design.

On the contrary - I think the most reasonable interpretation of his words is that he did use it via emailing it. But like I said at the beginning, I could be wrong. It will be interesting to see what he says when he returns to the conversation.

> Once again, I reiterate, an agent processing email that rejects every single one passes the test that the OP created, but then it can't do anything useful either.

No one is contesting that point, only that it is applicable.

> Author here. It was usable like any Openclaw agent. For example, I used it to ask it questions about the VPS, to summarize emails, etc.

It did not reject everything, it just stopped the costly processing.

> Is unwarranted.

Is this not a complaint?

But after a bit the cost grew so high that he just checked whether the attacks would have worked, without doing the costly response.

I could be wrong, of course, but it seems like the most likely interpretation of his words and why wouldn't be subject to your complaint.

(FULL DISCLOSURE - I used AI to fix some bad wording in my original version.)

My experience does not match yours.

And it's not just sometimes - it's most people.

So. They have their option too. And it's at the top, where most people can find it.

Understand that 99% are comfortable trusting downloads. They know that it's just as easy to sneak backdoors into source code as it is to sneak backdoors into executables.

See also: XZ hack.

I know ChatGPT seems like it's non-deterministic - but that just a user preference.

The only real issue is if you have many people on the same server, the GPU contention can be non-deterministic in rare cases.

LLMs can be deterministic if you need them to be - it's just that most people prefer the human-like interface.

Because then you can install it without depending on a package manager?

You think it's hard to obfuscate shell calls from inside a built executable?

What it tells us is that you're probably searching for reasons to grouse about AI.

Even when they implement LLMs.

You did not literally make that claim but your cost argument hinges on it.

Without it, then Claude does about the same as a human and only costs $100.

Apparently I'm reading your comments more thoroughly than you are.

I took the 10 most difficult patches from the git history - the ones that took the most back-and-forth to fix. I asked Claude to write them. Would you like to see the work?

If you believe a human performs better at finding downstream effects - you need to prove that. I see no reason why it should be true.

Claude Code doesn't need to be interested to work.

Let's see if they'll let this account through.