ctxc
ctxc on Hacker News
You're technically right, which is why I say "good enough".
It's a good strat imo - start simple and get it working with LLMs asap, then you can add heuristics etc based on the data you have and improve with that benchmark
What makes you say that? It's good enough
I'm not following, what did you mean?
Can you give me examples of things you don't like? Because I unironically admire how they improve UX. It's subtle, minimal, yet insanely useful.
Listed a few in my other comment
Won't disagree with you because I haven't used codex/OAI a lot - but Claude's non CLI UX is great too.
Eg - I asked for a recipe and it says leave x for y time - there's actually deeplinks that link to your clock timer with the time preset! And to change quantities, there's input boxes in the response that update all quantities accordingly.
Many such cases of use-case specific UX
I know this is an unpopular opinion, but it's actually good.
- pasting lots of content became [35 lines] and they added a tiny notification to show it as-is if you paste again
- interactive questions layout
- paste images
- the recap
- auto-updating session titles
And so many more. I tried codex and it was lacking in many UX aspects. Remember we got from a text-paragraph-only terminal to this at breakneck speed.
Claude code has great UX.
alright, where do I rsvp?
Posted by
ctxc on Saturday September 12, 2026 @01:53PM
from the self-post dept.
I have the automatic "co-authored by" Claude signature turned off for my commits via the config setting.PSA: They've introduced a new "Claude-Session: https://claude.ai/code/session_xyz" line that's appended to the commit now though, on by default.Annoying, feels scummy (of all things, ironic I know)
It's insane, right? Reverse engineering is so much fun, I find myself doing it on random apps so often now
agreed with "the final step". I worry adding commands like these might affect quality of responses at each step, which then stack up to result in an overall worse outcome.
Probably because they skew towards higher information density
"LemonLime bills itself as a no-code AI startup, not to be confused with the hundreds of startups building something similar."
Ouch
20 games, of 104? Not to mention games were played in indoor air conditioned stadiums.
MLS already has a system where a game has a break only when it reaches 28deg C for a period of x minutes
I didn't think about this parallel, but that's just so sad.
Thanks - I didn't verify this personally, but still stand corrected.
If training wasn't considered outside the law, this goes on to make the point about double standards for US vs Chinese model training methods.
Anthropic was asked to pay book authors because it trained on pirated downloaded books.
What about books and art where the author/artist does not authorise AI to train on it?
They do happily train on it, ignoring their "ToS".
This is just double standards, a slap on the wrist to not worsen the situation with authors imo.
You could also say the Chinese companies are doing the same - they _do_ pay for their Anthropic subscriptions after all.
Yeah weird, bc I just checked and it isn't oss. Only the SDKs are OSS, which...doesn't help much
/s not needed, it's the easiest path
(you mean it's _not_ AI generated)
I feel like with time, 50 is the new 70.
Those numbers make sense to me...120 million input tokens is like 120 sessions of hitting the full context limit, which seems like a lot to me though
Reads like a LOT of tokens to me. What does your usage /workflow look like? I'm v curious because although I do use Claude code, my token counts aren't nearly as much
I want to know if I'm missing something cool!
When you have a general idea of what smells bad vs what's okay...why?
I'd rather get it from the LLM and review
They report an hour less than average sleep time the first 3 months?!
How did get so lucky?