alt.hn

7/27/2026 at 11:33:04 PM

Interview with Boris Cherny [video]

 https://www.youtube.com/watch?v=qyPCVqFUyDo

by knighthacker

7/29/2026 at 8:02:02 AM

There’s something missing in the discussion. Building software is about achieving outcomes, not about generating output, ie tokens. Yet, the ai business case is selling output, obviously. Not sure how to reconcile the two. Ideas?

by DennisL123

7/29/2026 at 8:36:24 AM

Yes this 100%. I'd say: don't overrate how you do it; it matters more what you do, especially with LLM models. It's just a tool.

by zazuke

7/29/2026 at 12:57:22 PM

Building bridges is about achieving outcomes and utility, but steel and cement suppliers are selling by ton.

by yetihehe

7/29/2026 at 2:17:14 PM

To extend this analogy, vibe coding can be defined as dumping steel and cement out of a helicopter into the water and calling the result a bridge because you can technically walk on it.

by miyoji

7/28/2026 at 9:41:59 PM

I don't get why Anthropic keeps saying how cool it is that they were able to rewrite bun in rust, I feel like if the model is really that good you wouldn't need a rewrite, you'd be able to create the runtime from scratch.

by iooi

7/28/2026 at 10:27:43 PM

Code is a fantastic way to capture requirements, particularly iterative product discoveries over time. It's much easier to start with a manifestation of the requirements (and tests of invariants) than starting from base principles.

by chromakode

7/28/2026 at 9:52:02 PM

Productivity theatre is when activity becomes celebrated over outcomes. It's absolutely everywhere in US tech culture.

by __alexs

7/29/2026 at 5:28:42 AM

I mean if the model were really that good they'd be able to fix various TUI issues that have plagued claude code since last year.

by sushid

7/29/2026 at 5:32:33 AM

is flickering fully fixed yet?

by nnx

7/28/2026 at 9:48:33 PM

I like this, it will be a really exciting benchmark to refer back to in a few years

by jdross

7/29/2026 at 3:02:55 PM

it was partially a marketing stunt

by whateveracct

7/29/2026 at 12:04:55 AM

Rust is a great tool for what they did with Bun. Why make a whole new tool when existing off the shelf tools can do the job?

by schainks

7/28/2026 at 9:53:52 PM

I feel like the Big Insight with Claude Code was 'give the AI access to tools, in fact to your actual laptop'. And of course 1.5 years later 'a harness to enable an agent to use my computer' has turned into the flagship product we expect from frontier labs

However. I don't think all the other 'Ideas' these harness-building guys have are all that universally applicable. Heck they are naysaying their own gearhead stack of skills.md and claude.md lately including in this video. But also I'm suspicious of this whole thing of spawning agents. For example "find every function in this codebase" is probably better done deterministically using a script that extracts function names rather than by spawning 20 agents to 'read' chunks of the code in token space. But it certainly racks up the token usage which is good for the person selling you inference...

And there is a lot missing in the actual harnesses frankly that isn't about more parallel agent ninjutsu. Like why isn't there a 'move this function from this file to this file' tool (copy/paste char range) and we have to see Claude/Codex/etc flail around rewriting huge chunks of code in token space

by firasd

7/28/2026 at 10:51:49 PM

> Like why isn't there a 'move this function from this file to this file' tool (copy/paste char range)

There is, you can install LSPs for a given language which act as just another tool the model can use to more efficiently manipulate code.

by rescripting

7/28/2026 at 10:54:51 PM

LSPs don't really do this. They're like a better way to grep for something (so they do address the previous thing I said about listing all function names etc.) But they often don't say (depending on the particular LSP setup etc) this function ends at this char and they definitely don't then provide the text editing tool

by firasd

7/28/2026 at 11:42:36 PM

This is not accurate. Most LSPs do provide text editing tools, which the Language Server Protocol calls "code actions". These actions do things like: reorganizing imports, expanding macros, extracting a selection into a function, or converting one type of control flow into another (e.g. if to match). Common LSPs, like rust-analyzer and Ruff (Python) support code actions. Editors also expose code actions: Zed, for instance, exposes actions under the default keymap with the binding Command + .

source: https://microsoft.github.io/language-server-protocol/specifi... and https://microsoft.github.io/language-server-protocol/specifi...

by throw478239

7/28/2026 at 11:55:17 PM

Come on what are we talking about. Ruff is a linter. Can you show me any LSP server that can move an arbitrary function from file1.ts to file2.ts based on calling it on the command line?

by firasd

7/29/2026 at 12:16:00 AM

They're not command line applications, they use JSON-RPC inline, and it's an interactive flow, so you need the client to actually persist state.

So what you'd see in the LLM traces, for example, is a couple back-and-forths using JSON-RPC. It absolutely works at the moment, and claude code will happily use it if everything is set up correctly, just tested it using https://github.com/typescript-language-server/typescript-lan...

by jaggederest

7/29/2026 at 12:33:45 AM

What did you test exactly?

Look what I’m asking for is not complicated

Node cutpaste.js inputpath startcharnum endcharnum outputpath startcharnum

I was just using it as an example of how dropping claude / gpt in a Linux shell and saying good luck and then giving interviews about spawning sub agents seems to overlook basic text editing primitives

But if you managed to move an arbitrary char range from one file to another using Claude Code talking to the TS LSP server let me know that would be enlightening.

by firasd

7/29/2026 at 5:08:00 AM

The relevant points have already been made: (1) LSP is a JSON-RPC protocol, not a CLI app, and (2) Language Servers that implement LSP often utilize AST manipulation, even the resultant manipulations are exposed to the client via character ranges.

Still, these links may be of interest:

TypeScript’s language-service implementation of the “Move to file” refactor: https://github.com/microsoft/TypeScript/blob/main/src/servic...

Documentation for how `typescript-language-server` exposes that refactor as an opt-in code action: https://github.com/typescript-language-server/typescript-lan...

`ast-grep`, an AST-based search and transformation tool that may be closer in spirit to the cli functionality you describe, albeit more general-purpose: https://github.com/ast-grep/ast-grep

by henrymerrilees

7/29/2026 at 9:55:12 AM

Interesting. But I think this also proves my point that if this particular LSP server has the move statements functionality the pattern is not default integrated into Claude Code, OpenAI Codex, etc. Like you’d have to figure out a way to keep this server running yourself as if you were running Redis and asking the harness to manage your ToDos in it or something

by firasd

7/29/2026 at 10:00:18 AM

Jesus christ dude, what you're describing is what an LSP is. It is a server that runs, and integrates with editors, and claude code uses the LSP the same way an editor does. It does, in fact, stay running in the background along side claude code as an independent process. There's no "gotcha" here.

by jaggederest

7/29/2026 at 10:43:30 AM

Again this is exactly equivalent to saying Claude Code doesn’t need a todo tool (which it has ie TaskList) cause you can run Redis. What you have demonstrated is that there is a lengthy workaround depending on if a particular LSP server implements a move method .. to do text editing

by firasd

7/29/2026 at 4:27:58 AM

No, not an arbitrary character range, it's quite a lot smarter than that. It's an AST transformation - it takes the parsed tree out and puts it in another file and updates any imports that refer to the prior function to the new location. That all happens in the LSP side, not the LLM side, so it's deterministic. You can look at the source code for the LSP yourself.

by jaggederest

7/29/2026 at 12:14:38 AM

I've had decent results telling agents to use emacs for structural editing operations. Pretty rare that I need to move a function across files verbatim but that should only be a few extra lines of elisp.

by wild_egg

7/29/2026 at 1:43:06 AM

not sure if Go pls supports moving between files, but it does have some sophisticated edit features

https://go.dev/gopls/features/transformation

it does look to have a method for moving a function to a new file

I give my agent access to go pls and find it overall beneficial, based on vibes eval. Its way easier to ask LSP everywhere a function or field is used versus grep, saves a lot of tokens

by verdverm

7/29/2026 at 6:42:55 AM

> Like why isn't there a 'move this function from this file to this file' tool (copy/paste char range)

There Is: https://github.com/oraios/serena

by senand

7/29/2026 at 6:23:29 AM

Yes exactly, his job is first and foremost to be a token salesman.

He’d be telling you to use workflows that spin up loads of agents all the time whether or not he actually thought they were the best way to be productive

by ifwinterco

7/29/2026 at 8:00:24 AM

"find every function in this codebase" is rarely a thing imo. Why would you want to do that, for some statistics? In many cases, what you want to do is find "all classes that implement some interface" - of course depending on the language. And reasonable languages allow agents to find that reasonably efficiently ("implements Foo") - a bit slower than specialized interfaces, but in a way that don't require you to maintain a giant index.

And yes, agents writing chunks of code or scripts to accomplish tasks is already what they do, when they determine it's efficient.

>And there is a lot missing in the actual harnesses frankly that isn't about more parallel agent ninjutsu. Like why isn't there a 'move this function from this file to this file' tool (copy/paste char range) and we have to see Claude/Codex/etc flail around rewriting huge chunks of code in token space

I think if we had agent-specialized languages that you would modify by directly changing AST or some other structure it would be more efficient. But, otherwise, it's way more efficient to train on existing corpus - like PRs - which is the same reason that internal LLM thinking is not binharic, but regular English (or Chinese).

by KptMarchewa

7/29/2026 at 1:28:44 AM

The most interesting part is Boris' recommendation to delete your Claude.md every 6 months to see what the model is capable of. The link in the post sends you directly to that moment.

by knighthacker

7/29/2026 at 7:32:31 AM

It would have been nice to hear about how to be more efficient, task/token wise, especially with Opus 5.

Most users are on a subscription, and aren’t able to run dynamic workflows, spinning up 1000 agents or leaving a bun rewrite going for weeks. Don’t get me wrong… I like cake.

by discordance

7/29/2026 at 7:35:51 AM

Seems like something they should work on, not customers. Reinventing my workflow every six months because a company/product is lazy is not high on my list.

by bmitc

7/29/2026 at 10:05:16 AM

What I read from that is that there's a chance claude gets worse using a claude.md and that's a usability issue on their side.

by lorey

7/29/2026 at 1:23:57 AM

So maybe SKILL.md and CLAUDE.md should have model version pinned?

Somewhat like AWS CloudFormation templates start with something like TemplateVersion: 2017-10-01

But maybe that's old school thinking still.

by albert_e

7/29/2026 at 7:06:26 AM

They should just be symlinks to CONTRIBUTING.md and other docs humans are meant to read.

The whole idea behind LLM is they use the same language and tools as us, so we literally don't design anything specifically for them.

by crazylogger

7/29/2026 at 1:27:25 AM

I'm thinking a lot about this recently as I'm building AQ.dev (Multiplayer coding harness). My instinct is to keep it simple and not trying to do too many shenanigans with these files and let engineers maintain control.

by knighthacker

7/29/2026 at 1:39:26 AM

this assumes everyone uses the same model, but where I'm at, we use a wide variety of models

we keep the AGENTS.md simple and monorepo quirk focussed, no instructions for specific agents or models

by verdverm

7/27/2026 at 11:55:02 PM

The TL;DR of the fragment at around 6:57 in the video, for those who don't want to watch, is that Boris recommends to delete all the customizations, CLAUDE.md, skills, and other stuff you had for older models, and try using the new models without all that, because the new models can surprise you.

by ventana

7/28/2026 at 9:42:55 PM

All of these harnesses should support pinning config files and tools to specific models or families of models.

It's really tiring to have to tweak everything with each model release and then watch those changes mess up cheaper models in the process.

by m_ke

7/29/2026 at 1:47:01 AM

Yeah or config older than x days...

by swader999

7/29/2026 at 8:27:41 PM

What if I don't use customizations at all, bare prompt per feature or function only? Do those customizations help?

by ahmetson

7/28/2026 at 11:13:58 PM

Steering/corrective instructions are now supposed to go into memories - which are non-portable? How do we manage these documents for team members without everyone repeating themselves? And if you remove all these documents, how do you work with other models which might use them (nevermind that claude ignored AGENTS.md for the longest time).

I am on board with not putting stuff like "write clean code" into an agent file, or using plugins for tools that are now built into the harness. I don't see enough evidence to support models being significantly better at figuring out intent, or getting the assumption correct. I've always gotten better results (as ever) with very constrained instructions, vs "fix the install".

by throwaway219450

7/28/2026 at 11:26:51 PM

> Steering/corrective instructions are now supposed to go into memories - which are non-portable?

I’m sure this just intended to steer you to vendor lock-in. Remember, these are the same people who are so insecure/petty about their product that they won’t make it recognize the .agents/AGENTS.md standard.

by anon373839

7/29/2026 at 4:47:04 AM

The ability to start a Claude Code session in a temporary "config free" mode, similar to how incognito works for chrome, would be a nice feature to make this easy to try out, especially since model releases are so frequent.

by joebates

7/29/2026 at 6:21:04 AM

You can do this but you have to pass a million long options like --disable-system-prompt --allowed-tools "" etc

I have some wrappers but the only one I really use is a variant of that pinned to haiku for quick questions.

by matltc

7/29/2026 at 6:26:10 AM

`claude --bare --system-prompt ""` is a good starting point!

by kxxx

7/29/2026 at 4:39:44 PM

Yep! Only (potential) downside to --bare is it runs in print mode; can't kick off an interactive session with it. The flags I listed above are a close approximation.

Anthropic had something about API-only in the description a couple months back, but they've been waking all those back for months since the fable access rollback fallout

by matltc

7/29/2026 at 2:00:18 AM

There’s really not much to specialize in and differentiate yourself. All this harness customization is fleeting. Remember prompt engineering?

by sivanmz

7/29/2026 at 11:35:11 AM

To be fair prompts is just English language intent abstracted for code. It is still am very much a thing. Why do you feel that harness customization is fleeting, enterprises care about them a lot for many reasons from compliance to security.

by aiisjustanif

7/29/2026 at 12:04:46 AM

That is the interesting part to me. Also, was anyone else familiar with the LLM-app terms "ablation" and "product overhang," prior to this? I was not.

If you were, please share the knowledge with us. Where can I learn more?

by consumer451

7/29/2026 at 6:03:11 AM

Ablation is a medical term, he's just using it generically to describe removing existing config.

by iambenm

7/28/2026 at 9:54:52 PM

> because the new models can surprise you.

Yes, they keep surprising me that they still keep doing all of this: https://news.ycombinator.com/item?id=48962703 with no improvement despite all the marketing assurances that "hey you don't need to read code anymore"

by troupo

7/29/2026 at 5:04:31 PM

I do have to still remind codex to fetch large amounts of db records in batches otherwise it will always make a first false step of a naive select all style approach. Context wise naive can be ok for small datasets though so codex didn't know the naive approach was not viable. Context management the new code plumbing these days

by darepublic

7/29/2026 at 7:18:09 PM

Context management doesn't help. I had this requirement explicitly in my AGENTS.md file. Just a minute or so into a coding session it (and Claude) would ignore/forget half of AGENTS.md (because the advertised 1M token context is nothing but marketing bullshit)

by troupo

7/28/2026 at 10:50:30 PM

Yeah, I'm going over a bunch of components a frontier model generated and while they "work" the code is quite shocking and will create significant maintenance burden. Not to mention they don't use any of our shared utilities and duplicate so much code.

by slopinthebag

7/28/2026 at 9:27:09 PM

Even for older models I was (and am) of the opinion that a lot of the context-overstuffing cruft people have been wrapping around their LLM usage is more of an RNG-impacting (but not always for the better) lucky charm rather than it being universally helpful.

And a lot of people latched on to it as a form of self-soothing.

"I may not write much code anymore, but I can still be an expert prompt engineer!"

by georgemcbay

7/29/2026 at 12:23:45 AM

Not really. The big change in newer models is the amount of RL relative to pretraining. RL makes the models stable by collapsing distributions around desired outcomes. The old tweaks like giving the agent a role ("You are an expert XYZ engineer") or stuffing the context with engineering idioms have effectively been RL'd into negligence, but they still waste tokens.

You can see this visually in older image generation models. Stable Diffusion 1.5 would produce wildly different images based on slight variations in prompt and seed, but the latest image gen models are nearly seed indifferent and can tolerate a decent amount of prompt tweaking while staying "consistent"

by CuriouslyC

7/28/2026 at 9:59:23 PM

> is more of an RNG-impacting (but not always for the better) lucky charm rather than it being universally helpful.

I wrote this last year, it's still true:

--- start quote ---

https://dmitriid.com/prompting-llms-is-not-engineering

In reality these are just shamanic rituals with outcomes based on faith, fear, or excitement. Engineering it is not.

--- end quote ---

There are some prompts useful for the user like brainstorming [1] but on the whole it's nothing but lucky charms

[1] brainstorming from superpowers: https://github.com/obra/superpowers

by troupo

7/28/2026 at 9:06:41 PM

CHANGELOG: surprise me

by ares623

7/29/2026 at 1:18:43 AM

Yup. That's pretty much the most interesting part.

by knighthacker

7/28/2026 at 9:53:53 PM

Salesman "recommends" you to continue wasting more tokens and on top of that, he "recommends" you to use only recursive loops with the latest and greatest models.

Finally he also "recommends" that you do not look at the code, or even understand it.

His "recommendations" are designed to get you to spend even more tokens and get you hooked on the Opus / Fable slot machine in order to extract as much money as possible from your wallets.

News at 10.

by rvz

7/29/2026 at 7:06:10 AM

I'm waiting for the next headline that's about a codebase that existed for 10+ years and then the agents came.

Now, the codebase is managed by agents and, while the company has the source, it's as if they bought it from a vendor and pay the vendor for changes. The vendor of their own system is the AI company selling them tokens.

by hvb2

7/29/2026 at 11:18:52 AM

To be fair I think Cherny is high on his own supply, not cynically trying to sell us something. He is absolutely stoked that Claude Code is an overengineered piece of junk he doesn't understand.

by Diogenesian

7/29/2026 at 7:26:28 PM

Boris Cherny is most definitely the kind of guy who sniffs his own farts.

He can be informative to listen to, as long as you keep that in mind.

by KludgeShySir

7/29/2026 at 1:15:06 AM

I was using Opus 4.6 until 2 days ago with no CLAUDE.md or anything and it was great.

Tried out Opus 5 and it's been a super annoying experience out of the box.

by mupuff1234

7/29/2026 at 3:34:36 AM

> From what I'm hearing, you for every model release, you basically delete all of the codebase, delete all of the prompt and start from scratch every time. That in the old world would have been not something Startup would have done for the product. It's like press delete every six months for everything.

> That's right. So to be fair, we don't delete the entire codebase.

by nojs

7/28/2026 at 10:26:48 PM

I've spent the last year building an app with Claude Code without knowing how to code. I'm still surprised at how much it gets wrong, even on the newest models. What makes it work for me is describing exactly what I want and then actually checking everything it gives back.

by mfallon

7/28/2026 at 11:09:00 PM

For me, I just don't trust Claude Code to write first-class professional code. I have used all frontier claude models and it always does sloppy job that needs to be fixed by codex. I have claude/codex setup and any execution must go to codex and all reviews as well.

by EagleEdge

7/29/2026 at 3:21:35 AM

Similar here. I use Claude Code and Codex for developing specs, but then have Deepseek v4 Pro implement because it just seems to actually do what it's told, given enough tries. And Reasonix + Deepseek is mostly cache hits when I let it grind on it for a long time, and I can have a huge fleet of agents with very specific focus for code review and rule enforcement mostly sharing cache.

by wilj

7/29/2026 at 12:27:13 AM

2 weeks rebuilding in Swift is absolutely crazy. Wonder if they're using production Opus 5 for this or their internal Mythos (if they even use it internally)

by jke_kang

7/29/2026 at 12:38:13 AM

If it's on the house (and they can probably turn up the token speed) why use anything other than the most powerful models?

by kev009

7/29/2026 at 5:09:18 AM

it's still more expensive and slower to run bigger models though. resource usage is not free

by blovescoffee

7/29/2026 at 4:23:00 AM

still a long way to go!

by linxy97

7/27/2026 at 11:42:58 PM

Never create it in the first place

by lostmsu

7/29/2026 at 12:54:54 AM

30:08 "This means engineers can talk to users... stuff that's actually fun". Wow, he totally gets us. /s

31:09 He flubs the questioning. His first says "Raise your hand if 100% of your code uses agents." Then he says, "What about more than 50%". You can see the same audience members who raised the first time shrug and also raise the second time. He then says "Slightly less hands," which just means he sees what he wants to see. Don't believe this man.

by jake_and_fatman