I read that as "you really don't" need rich friends to buy a H200. I see what you meant now. Sorry about that. The shit you hear people say in SV sometimes is wild.
I run them on my 5070. They're moderately useful at that scale I guess.
The only difference between CLAUDE.md and AGENTS.md is the filename. Do you really need a whole plugin system to support that use case? Feels like this could have been a one line change.
Don't you run bizantine ralph loops on remote environments with codex security checks, coupled with jev, grok, open router and a fully independent openclaw (on a maxed out mac mini inside a caveau in an undisclosed location, with open telegram) to change constants? You're going to be left behind.
Not related but I think I finally know how it must have felt to the generation prior to mine, when we tried to explain computers to them; because even as a retired 30 year plus software engineer, I have no idea what the comment says.
I felt a little panic when I understood every word of what was meant to be adjacent to scifi-style garbled engineering speak, and I think I'd rather be on your side of the spectrum.
It's almost all related to LLM's, which is fairly new.
Basically he's describing an LLM assistant setup that does coding on a vps or Mac mini ( some contradiction with Openrouter though) where you can communicate with the assistant through telegram
OK I do use AI in the browser but haven't tried anything like what you describe. I Googled telegram in case it wasn't what I was thinking, which was used in the 1800's but found their website. It's apparently another messaging platform. Thanks for the interesting view.
The tl;dr of the tl;dr is treating the model setup like an autonomous robot with its own machine to use and just telling it what to do via chat rather than directly supervising it, which is bonkers in some ways (all the open internet/service access/permissions stuff you would imagine), and smart in others (a fully isolated machine means it can fuck up something locally or run high-load tests without taking down your important systems).
Unc you're so cooked it's not even funny younger than me but you're out here aura farming in the negatives fr fr. NPC-ahh behavior, 0 aura, -1000 rizz. Couldn't be me
Sorry couldn't resist ( ・ั ﹏ ・ั )
Not sure what you didn't understood in their comment though, maybe the raph loop (and jev which was just released)? Isn't the rest completely self explanatory?
The plugin system itself was probably already in the making, and they just chose to implement this tiny feature as a plugin to try it out.
As for the only difference being the file name, that's an untested assumption. Up until now, Claude hadn't supported AGENTS.md, and it's a simple application of Hyrum's Law that somebody, somewhere, was taking advantage of that to give one set of instructions to Claude, and a different set of instructions to some other provider. The correct behaviour in the presence of both files is not obvious, either.
Having a kill switch for the change is a perfectly reasonable safety measure in case something goes horribly wrong.
> somebody, somewhere, was taking advantage of that to give one set of instructions to Claude, and a different set of instructions to some other provider.
I'm this person, I am actively doing this! I have some instructions which the Claude models need and OpenAI models will do by themselves. If the OpenAI models see the Claude instructions, they will go way overboard in a way I don't want.
Mind, the change does not affect me at all, because when both instructions are present Claude continues to read CLAUDE.md and Codex continues to read AGENTS.md. But if Claude started preferring AGENTS.md, that would be mildly annoying.
Within agents.md, an idea. Feel free to steal if you work on this:
Model.Claude*:
Do this one thing
Model.Claude.Opus.4.8:
Extra specific instructions
Model.*:
This is relevant to everyone
Harness.ClaudeCode*:
Why not, let’s allow harness specific instructions so you can do things differently depending how you’re working
Seriously, though, this is a great idea. Good communicators know to tailor their discourse to their audience. LLMs are just as sensitive to small variations in prompts, and just as inconsistent between them, as are humans.
I wrote an agent launcher sort of bash script. Pass in the command to start the agent, the script checks if there’s a Jinja file in a special directory matching that name, and builds it to AGENTS.md. Then it launches the agent.
I was trying to use it as a sort of janky RAG. I had a bunch of snippets (one for DB architecture, one for how load balancing works, etc), and my Jinja files were mostly a list of snippets to pull in. Voila, a bunch of agents that share little pieces of info but have a single source of truth.
I never got a ton of value tbh, it was very good at just grepping the snippets.
This "tiny feature" consists of hundreds of lines of code spread across 57 source files, many of which are less than 30 lines long, often contain more comments than code, and have no real reason to exist. It's the very definition of overengineering and overcomplexity.
That said, I do find it somewhat funny that Anthropic themselves are falling victim to their own model's tokenmaxxing.
How, exactly, would the proposed solution (combined with a setting to disable it) break, and how would implementing it the way Anthropic did address that? Be specific.
Some people might have a different CLAUDE.md and AGENTS.md because they use Claude to do one thing and $other_agent to do another thing. Or because Claude does XYZ by default, and other agents do ABC, and they want to give the correct guidance to both.
This is, to me, a completely reasonable and believable use case that could break if you implement this fix without "overengineering" it.
A setting to disable what? Be specific. See? Not so simple...
First, GP's proposal already addresses that. If both are present, CLAUDE.md would be used. Second, that is solved with a settings toggle. Read a boolean from .claude/settings.json and disable the new behavior if it's true (or false, depending on what you want to name the setting). third, you skipped the second part of my question: "how would implementing it the way Anthropic did address that?" Implementing the same behavior through multiple layers of abstraction and an order of magnitude more code doesn't solve the issue you mentioned.
Bonus forth point: why is this critical to solve for claude code, but not for all the other harnesses which have all converged on AGENTS.md for this purpose?
Okay, so we agree that this fix isn't quite as simple as it sounds then, yes? We've just had 3+ paragraphs of discussion around potential edge cases and additional considerations beyond "read one of two files."
We've had a long discussion only because you refuse to admit that a very simple solution would work (despite completely failing to show how it wouldn't). The problem is that your opinion is not in fact proof that you're right.
I see, well I'm not particularly concerned with being right on this topic (call it a difference of opinion or matter of taste), so I wish you a pleasant day.
Which order do you prefer them in? Since CLAUDE.md is no longer necessary, but was for so long, what happens when users only update AGENTS.md going forward while keeping a stale, unchanged CLAUDE.md around?
We're talking about changing default behaviors here in ways that can be surprising to users. It's reasonable to try and accommodate existing setups, future setups, and - yes - careless users.
There's no surprising behaviors in that situation. Someone who is aware of the change would most likely delete CLAUDE.md. Someone who is unaware of the change would assume CLAUDE.md is still the one being read by CC, which would hold true.
Is accusing someone of "shilling for Anthropic" for suggesting that a bug fix might not be so simple really the level of discourse HN has devolved to now? Is it the kind of HN you want?
Is the level of HN discourse suggesting that having a large number of users worth of never making simple changes something you want? This isn't a move fast break things suggestion, this is a rational common sense suggestion. Instead, you're pushing an overly complex way of implementing something that would allow for all sorts of unknowns. Earlier comments suggesting this new mods concept was being worked on giving the dev a new 20lb sledge hammer so the CLAUDE/AGENTS situation suddenly got beat on like it was a railroad spike when it just needed a smaller hammer since it was only a finishing nail.
This conversation comes down to people saying “Why don’t you just…” to Anthropic. Anthropic has the most information and made the choice they made. That choice may have been over-engineering, an appropriate choice or both. Discussing that choice could be interesting and it could especially be informative to people without much experience. Speculating on that could lead to an interesting conversation—especially WRT how people are actually using the current status quo—but imagination seems to be lacking here.
Instead we get the implicated assertion that users’ existing workflows shouldn’t matter to Anthropic and accusations that someone you don’t agree with is a shill. When called on that you avoided the question.
To answer your question, although it wasn’t directed at me: YES. I want people to be able to discuss that exact issue even if it isn’t personally interesting to me.
> Anthropic has the most information and made the choice they made.
Anthropic is the last company I would trust to make any decisions. Look at any discussions surrounding their "Claude is a tiny game engine" idiocy, numerous bugs that a junior can discover, a full "plugin system" in which they neeeed a dozen files in the worst Clean Code manner to read one of two files etc.
I've never heard of the 'move fast, break things' mentality ever giving a damn about number of users. If so, no changes would ever be made. The great thing about having millions of users is QA can be eliminated entirely as you'll start hearing about issues from the users directly.
Bash does this well, reading only the first of half a dozen config file locations it searches. And if the user needs it to read from an additional location, just add a source line to that location.
> Having a kill switch for the change is a perfectly reasonable safety measure in case something goes horribly wrong.
What, exactly, is going to go horribly wrong here ...? The worst case you're proposing is that an Anthropic model acts on AGENTS.md rather than CLAUDE.md. It's not going to be the end of the world; at worst it'll be `git reset --hard HEAD` time. But this wouldn't happen, because:
> The correct behaviour in the presence of both files is not obvious, either.
It seems extremely obvious to me: the non-breaking change is the right change here. If both exist, Anthropic models continue to use CLAUDE.md and ignore AGENTS.md. This is not rocket science. Why are we overthinking this?
> What, exactly, is going to go horribly wrong here ...?
Operating anything at scale teaches you that the universe of “things that can go horribly wrong” has a much better imagination than you. That goes quadruple with complex and squishy reasoning models.
I’ve seen a couple of really good examples of what can go wrong in this subthread, the most straightforward of which is that abruptly changing a fundamental behavior can break existing use cases.
Because when you have tons of users ainor fuckuo is a big fuckup and also it's really common to have both claude.md and agents.md and use @ syntax (which lets you reference markdown files when using Claude code, but not other harnesses) so you Claude md looks like
```md
@AGENTS.md
[Claude specific stuff]
```
And then what happens if someone now puts @syntax in their agents.md triggering a loop etc. It's all vibe coded - including code from days with dumber models - there's gonna be all sorts of dragons under the hood
The Claude sandbox has a slightly different ENV compared to ChatGPT (LC_CTYPE set to C instead of UTF8 I believe) which upsets some scripts we have.
So in the CLAUDE.MD we tell it about that and then tell it continue reading AGENTS.MD. Not very hard to work around if CLAUDE.MD goes away, but just to support your point that the split did have its uses for some.
> Having a kill switch for the change is a perfectly reasonable safety measure in case something goes horribly wrong.
Having a way for the user to revert the update is a perfectly reasonable safety measure in case something goes horribly wrong.
Even though Anthropic or any other software provider could technically run anything on our system, there exists a modicum of trust between supplier and user. Anthropic performing arbitrary changes that are tangential but not specific to the premise of delivering a harness to package code files and prompts in context to their LLM breaks that trust for many users.
I'd love to hear more about this. Though I usually write bespoke software for specific clients, I'm about to release my first general-availability application. It's a rust application with Python GUI on the desktop (Linux, Mac, Windows) and a skinny Android application that wraps the rust on mobile. I'm going to be gating features starting probably next week. I'd love to hear tips and experience for implementing feature flags, for both the desktop and mobile space.
> The only difference between CLAUDE.md and AGENTS.md is the filename. Do you really need a whole plugin system to support that use case? Feels like this could have been a one line change.
I don't think this is a reasonable assumption. The document format in CLAUDE.md is whatever Anthropic specifies, where AGENTS.md is a common ground format that is expected to be supported by any agent, be it from Anthropic or not.
I think you're giving these files too much credit. There are no specs, they are freeform text. In this sense they are the same. The expectation of what could be in it by each vendor means nothing unless it's enforced.
There is no "specification", it's literally just a text file that gets loaded into context. You don't even need to write markdown. It's not that hard man.
I'm fairly happy with uBlock Origin Lite. Sure the original was better but this one does the job as well. uBlock Origin Lite is a MV3 extension, by the same author.
https://github.com/uBlockOrigin/uBOL-home/wiki/Frequently-as... has a few of the more practical concrete examples. Namely content can load without the blocker being ready to filter, CNAME cloaking can't be handled properly (in Chrom* at least), and the filter rules aren't as flexible.
Interestingly, the rule count limits themselves do not (currently) seem to be much an issue despite it being one of the more common concerns on the original MV3 announcement.
> content can load without the blocker being ready to filter
This isn't an MV3 issue, it's a Chromium design decision that extensions cannot block browser initialization. It was also true with MV2. In fact, MV3 has improved the situation because DNR rules are enforced immediately.
You could certainly argue that Chrome should allow extensions to block browser initialization, or that it should be user configurable - but it seems like an entirely defensible Chrome decision. Regardless, it's nothing to do with MV3.
> the filter rules aren't as flexible.
That's true, but they are still very flexible, and DNR makes some very nice features possible in uBO lite that were not possible in uBO.
For example, did you know that you can change your User-Agent with uBO lite? Go to the Custom DNR rules tab, and enter something like this to pretend to be Lynx on whatever.com:
Any MV3 issue could probably be summed up as "MV3 isn't so bad, just the implementation in Chrome" where Chrome defines MV3 (as it's not a W3C standard) and ships it as such.
That said,
> Chromium design decision that extensions cannot block browser initialization
It's not "browser initialization" that was blocked, it was blocking network requests for pages until they were processed by all components in the chain. Notably, this could never apply applied to the browser or its internal pages itself.
Firefox is a great example about all of the above points: MV3 still supports blockingWebRequest there, so it gets the best of both worlds.
> did you know that you can change your User-Agent with uBO lite?
With MV2 you'd use blocking webRequest and chrome.webRequest.onBeforeSendHenders. I'm not sure uBO ever bothered with this before uBO Lite, but it's a bit of a double standard to say "MV2 didn't offer blocking requests, that was a different API which happ
> and then the request had to wait until it was given to and processed by the blocker.
Despite the name, the requests were not blocked until the extension had initialized. The browser could make network requests before that happened with MV2. With MV3, static rulesets and persistent DNR rules are enforced immediately.
> With MV2 you'd use blocking webRequest and chrome.webRequest.onBeforeSendHenders
Of course an extensions could do this, I said uBO lite. DNR made exposing this powerful functionality fast and easy. My point is that MV3 uBO lite has useful features that MV2 uBO does not.
> Despite the name, the requests were not blocked until the extension had initialized. The browser could make network requests before that happened with MV2. With MV3, static rulesets and persistent DNR rules are enforced immediately.
There were several workarounds like suspendTabsUntilReady which helped greatly in this half of the problem. Then it would solidly stay permanently registered too. I'm pretty sure none of these work in Chrome anymore though, it was obvious they never intended these sorts of things to work that way.
Nevertheless, MV3's ability to offer limited filtering on startup does not seem like such a blessing when the other option was to just guarantee full filtering on startup as seen in this change https://bugzilla.mozilla.org/show_bug.cgi?id=1447551
> Of course an extensions could do this, I said uBO lite. DNR made exposing this powerful functionality fast and easy. My point is that MV3 uBO lite has useful features that MV2 uBO does not.
What you said is so, it just might highlight more than what you'd like it to at the same time. That it was missing before had nothing to do with the difficulty or need for MV3. uMatrix had it for a while and Gorhill regretted it because it didn't change things like the user agent JS would report and such, hence he wouldn't port it over. The above filter has the same problems. Great if you just want a normal quality user agent switcher, that it's normal quality is why it was never in the old extension though.
> Nevertheless, MV3's ability to offer limited filtering on startup does not seem like such a blessing when the other option was to just support full filtering on startup
Sure, but as I said, it's a design decision, and nothing to do with MV3. I'm not trying to litigate it one way or the other.
I do think DNR is a good solution, and that request callbacks were a bad idea. That's my opinion - and the counterargument is obvious -- you can't do literally everything you could do in a callback declaratively. That doesn't sway me, I still think it's better!
> Great if you just want a normal quality user agent switcher
It was just a trivial example of using a powerful new feature that uBO doesn't have, not a full featured extension! The fact that it's possible at all is the point, it's cool.
You can definitely do anything that DNR can do with webRequest, nobody can dispute that. The point is that there are things you can do with uBO lite that you can't do with uBO (and vice versa!). It's not a straight downgrade, there are powerful new features.
Again, we really need to separate "MV3 - the theoretical idea" and "MV3 - the actual implementation which ships with Chrome". It's two confusingly different worlds.
Nobody should be swaying you to prefer DNR over webRequest (or vice versa) when other browsers show both can be available options for the extensions you use as makes sense for what they try to do/what type you'd like to install. Continuing to support MV2 power levels of functionality does not mean uBO Lite type extensions would need to stop working the way they do, it means uBO type extensions would continue to work the way they do. I.e. the biggest failure of MV3 is not in lacking anything new whatsoever, it's in forcing the removal of a lot of API options which were useful in the name of ulterior motives.
> You can definitely do anything that DNR can do with webRequest, nobody can dispute that. The point is that there are things you can do with uBO lite that you can't do with uBO (and vice versa!). It's not a straight downgrade, there are powerful new features.
It's like you sold your house for a mobile home but tell the neighbor it's not a straight downgrade because it's red. Great, but why are we praising that instead of painting the house red?
Funnily enough, this morning I ran across another feature I didn't realize was missing in MV3 https://github.com/uBlockOrigin/uBOL-home/discussions/369. Perhaps keeping the functionality to know what was actually blocked to the extension was too powerful.
> the biggest failure of MV3 is not in lacking anything new whatsoever, it's in forcing the removal of a lot of API options which were useful in the name of ulterior motives.
What was the ulterior motive? If it was to break ad blocking, it seems like they did a really terrible job, as MV3 ad blockers work great. That's not me saying that, here is a paper from researchers at Goethe University Frankfurt who tested it:
> Moreover, cross-browser experiments yield comparable outcomes, and visual inspection confirms that MV3 ad blockers work effectively without significant ad flickering or loss of functionality
> Funnily enough, this morning I ran across another feature I didn't realize was missing in MV3
I think you could accomplish this by redirecting instead of blocking, then increment a counter in a webRequest callback. There would obviously be a performance cost, as blocking is so cheap and callbacks are expensive...
Regardless, I think it's jarring hearing you call custom DNR rules "painting the house red", a legitimately powerful feature... and then point to a missing counter in the UI as evidence of lacking features!
> content can load without the blocker being ready to
That issue does not exist for uBO Lite. I specifically wrote in the FAQ under "What are the upsides of an MV3-based content blocker?":
> For Chromium-based browsers specifically, MV3-based content blockers which are entirely declarative will properly filter at browser launch, which is not the case for MV2-based content blockers
uBO Lite is entirely declarative, so it filters properly at browser launch, and I literally give an actual example of this in the following section, search for "filters properly at browser launch".
The companies that don't want you using the lite version are waiting for full version to completely go away before blocking the lite one. No sense driving you to that. So, check back in a couple months and see if it's all still good.
I've seen some ads slip through the cracks nowadays, like I think Wikia ads for some TV shows I go to now have ads on them. They gave up on blocking Meta.
I believe the main things are that block lists have to be updated through the Chrome Web Store and there is an upper limit for things that can be blocked.
The main reason uBlock on Firefox works better is because of CNAME de-cloaking. All the tracking shit that appears to load as first-party, but isn't, can be correctly detected and blocked on Firefox. On Chrome it cannot, and with Google, Facebook, et al. encouraging site owners to move to these fake first-party conversion tracking solutions, your exposure to profiling has increased dramatically. You may see fewer ads but it does not do the same job.
Isn’t the idea that over time chrome will limit what ubo lite blocks? It will phase out all of its own ad networks so they’re not blocked first, but it’s possible they also start allowing third party ads through for a price.
reply