comboy

Just show us your face and transactions history, it's about the kids.

You would not last long in China ;)

(you pay by scanning QR code in .. well, everywhere)

I agree. The intent is sacred. This should be the default and CLIs should make use of the available history (while preserving inputs you need to preserve outputs too because context matters).

The idea of having to repeat something to your computer is ridiculous.

methinks if you really want to have it versioned it should be AGENTS.md in VCS and your globally gitignored CLAUDE.md just @AGENTS.md

otherwise it's like leaving vim dotfiles in the repo or something

How do you envision short term and long term target usage of it?

And do you guys communicate between other browsers when doing something like this to try to settle on something common? I don't mean W3C but practically, it's a small world after all.

Code changes are cheaper to make now and kind of more expensive to verify.

So you can still contribute, you just not need to provide the code, just the issue.

Which isn't as bad as it sounds, it kind of feels bad to rewrite somebody's code right away when it is theoretically correct, but opinionated codebases seem to work very well if the maintainer opinions are sane.

Mostly datasets but who knows if LLMs stuff also if I'm creating a datasets with them[0]. Long story short it's a potential headache, don't want that headache for now. Plus basically two options after open sourcing: people are not using it (time and effort wasted) or people are using it and then it's a chore. But still on the table. But I'm currently not close to the table.

0. https://claude.ai/share/23d27617-7608-4c58-93b9-b9613bc29318

Only about 5K explanations now I'm still trying to polish the pipeline before covering more. Due to all these verification and associated regenerations the cost is quite high.

It's not opon source but it is completely free. Open sourcing is on the table but currently it would be additional work and distraction. Just licenses seem like a headache, from a quick poke even when ok for use on non-commercial data redistributing may not be ok. So not any time soon, but if you are working on something similar I can at least share detailed datasources, finding good ones was not easy and LLM integrate them fast.

Great minds think alike :D https://hanzirama.com/character/%E5%AD%A6

It is also allowing me to see all relevant associations easily when revealing the card in built in SRS, you add cards to SRS as you browse, so they are related to what you already know / currently exploring.

Mind you, all data visible is collected from different reputable available sources. When you click "explain" there's a clearly marked LLM explanation, but my explanation generation pipeline pushed all generated explanations through 5 different models including all top Chinese-first for verification, and on average it took a few iterations back and forth to iron out any information that could potentially mislead the learner.

You can actually see thousands of words I typed just working on that pipeline here https://hanzirama.com/making-of

From my experience, saying "this is not X, it will be not used for Y" is vastly increasing chances of this being classified as being X. Anybody can write "this is authorized research". Instead use something like evaluate security / verify security, make sure this cannot be (...), etc.

Of course these models are pretty smart so even Anthropic's simple instructions not to provide any exploits stick better and better.

Tried antigravity and cli a few times. I'm unable to handle that prodigious toddler. Are you guys able to make use of it?

I mean, yes but LLMs have been making me more cognitively active. I've learned how to do more stuff that I would have without them and it's a decent multiplier not some rounding error.

Obviously you can have a plumber that knows his stuff and the one that doesn't. The good one can check some details and will recognize bs. If you already have the bad one it's probably if better if he uses LLM rather than when doesnt.

I think API is fine, likely only subscription is affected. Not to mention trivial heuristics to differentiate repeated API calls / same data and potential CLI usage although that would be true malice.

It seemed to me that it was performing better through opencode using API but did not test extensively.

Pretty reassuring to hear that. I was skeptical too, there's a lot of variables like some crap added to memory specific skill or custom instructions interfering with the workflow and what not. But now it was like a toddler that consumes money when talking.

We are sorry, but your print resembles random princess from Disney too much (actually, we won't tell you which). Just following the law you know..

Unrelated, but Claude was performing so tragically last few days, maybe week(s), but days mostly, that I had to reluctantly switch. Reluctantly because I enjoy it. Even the most basic stuff, like most python scripts it has to rerun because of some syntax error.

The new reality of coding took away one of the best things for me - that the computer always just does what it is told to do. If the results are wrong it means I'm wrong, I made a bug and I can debug it. Here.. I'm not a hater, it's a powerful tool, but.. it's different.

I have the same bias as the parent. I'd rather pay $50 one time than $9 a year even if I throw it away after 4 years.

But the main reason I wouldn't install it despite being happy customizing linux is that it's yet another black box I need to trust and that knows way too much. It's really insane how much you need to compromise your security on macos to have a decent developer experience.

To be fair, I can do 3 balls effortlessly, but I can't do 1 ball like it is in this description, I just have a lot of error correction, enough to do it pretty much indefinitely. But I cannot reliably throw it accurately to the other hand.

Our software stack is the opposite of that.

You just convinced me to try it. Claude just copy pastes, does search and replace, zero abstractions and I'm the one that needs to think about the edge cases.

Any good reasonable alternatives? Gemini is like prodigious 3yo hopeless for my projects, anybody tested some opencode with kimi or something?

Interesting that in Chinese, classical writing is associated with terseness. That definitely wouldn't work that way in other languages I know.

It's not a well-written book. It's an interesting book (more like a story).

Program defines the exact computer instructions. Most of the time you don't care about that level of detail. You just have some intent and some constraints.

Say "I want HN client for mobile", "must notify me about comments", you see it and you add "should support dark mode". Can you see how that is much less than anything in any programming language?

The spec manually crafted the user is ideal.

It's just that we're lazy. After being able to chat, I don't see people going back. You can't just paste some error into the specs, you can't paste it image and say it make it look more like this. Plus however well designed the spec, something like "actually make it always wait for the user feedback" can trigger changes in many places (even for the sake of removing contradictions).

Hey, you seem to have similar view on this. I know ideas are cheap but hear me out:

You talk with agent A it only modifies this spec, you still chat and can say "make it prettier" but that agent only modifies the spec, the spec could also separate "explicit" from "inferred".

And of course agent B which builds only sees the spec.

User actually can care about diffs generated by agent A again, because nobody wants to verify diffs on agents generated code full of repetition and created by search and replace. I believe if somebody implements this right it will be the way things are done.

And of course with better models spec can be used to actually meaningfully improve the product.

Long story short what industry misses currently and what you seem to be understanding is that intent is sacred. It should be always stored, preferably verbatim and always with relevant context ("yes exactly" is obviously not enough). Current generation of LLMs can already handle all that. It would mean like 2-3x cost but seem so much worth it (and the cost on the long run could likely go below 1x given typical workflows and repetitions)

GPUs can do graphics too?

It's hard to tell how much it says about difficulty of harnessing vs how much it says about difficulty of maintaining a clean and not bloated codebase when coding with AI.

I mean, tools change, but I'd be happy to hear if any tool can create that by just saying create "Claude Code Unpack" with nice graphics. or some other single prompt. It likely was an iterative process and it would be lovely if more people started sharing that, because the process itself is also very interesting.

I've created some chinese characters learning website and I took me typing 1/3 of LoTR to get there[1]. I would have typed like 1% of that writing code directly. It is a different process, but it still needs some direction.

1. https://hanzirama.com/making-of

As things stand today even when doing research tasks, time spent by model is >> than fetching websites. I don't see it changing any time soon, except when some deals happen behind the scenes where agents get to access CF guarded resources that normally get blocked from automated access.

Add CI to check if new laws don't contradict with any existing ones.