For the best experience on desktop, install the Chrome extension to track your reading on news.ycombinator.com
Hacker Newsnew | past | comments | ask | show | jobs | submit | history | summarity's commentsregister

Not claude code specific, but I've been noticing this on Opus 4.6 models through Copilot and others as well. Whenever the phrase "simplest fix" appears, it's time to pull the emergency break. This has gotten much, much worse over the past few weeks. It will produce completely useless code, knowingly (because up to that phrase the reasoning was correct) breaking things.

Today another thing started happening which are phrases like "I've been burning too many tokens" or "this has taken too many turns". Which ironically takes more tokens of custom instructions to override.

Also claude itself is partially down right now (Arp 6, 6pm CEST): https://status.claude.com/


Ive been noticing something similar recently. If somethings not working out itll be like "Ok this isnt working out, lets just switch to doing this other thing instead you explicitly said not to do".

For example I wanted to get VNC working with PopOS Cosmic and itll be like ah its ok well just install sway and thatll work!


Experienced this -- was repeatedly directing CC to use Claude in Chrome extension to interact with a webpage and it was repeatedly invoking Playwright MCP instead.

I actually submitted an upstream patch for Cosmic-Comp thanks to Claude on Saturday. I wanted to play Guild Wars remake and something was going on with the mouse and moving the camera. We had it fixed in no time and now shit is working great.

It’s as if it gives up, I respond keep going with original plan, you can do it champ!

[flagged]



They're saying just do it yourself instead of trying to herd an unpredictable animal to your bidding like an LLM.

Yes, and over the last few weeks I have noticed that on long-context discussions Opus 4.6e does its best to encourage me to call it a day and wrap it up; repeatedly. Mother Anthropic is giving preprompts to Claude to terminate early and in my case always prematurely.

I've noticed this as well. "Now you should stop X and go do Y" is a phrase I see repeated a lot. Claude seems primed to instruct me to stop using it.

as someone who uses deepseek, glm and kimi models exclusively, an llm telling me what to do is just off the wall

glm and kimi in particular, they can't stop writing... seriously very eager to please. always finishing with fireworks emoji and saying how pleased it is with the test working.

i have to say to write less documentation and simplify their code.


LLMs are next token predictors. Outputting tokens is what they do, and the natural steady-state for them is an infinite loop of endlessly generated tokens.

You need to train them on a special "stop token" to get them to act more human. (Whether explicitly in post-training or with system prompt hacks.)

This isn't a general solution to the problem and likely there will never be one.


Try Codex, it's a breath of fresh air in that regard, tries to do as much as it can.

> Whenever the phrase "simplest fix" appears, it's time to pull the emergency break.

Second! In CLAUDE.md, I have a full section NOT to ever do this, and how to ACTUALLY fix something.

This has helped enormously.


Any chance you could share those sections of your claude file? I've been using Claude a bit lately but mostly with manual changes, not got much in the way of the claude file yet and interested in how to improve it


Typo - "proove". "Prove" only has one O.

Thank you!

I switched from Cursor to Claude because the limits are so much higher but I see Anthropic playing a lot more games to limit token use

What wording do you use for this, if you don't mind? This thread is a revelation, I have sworn that I've seen it do this "wait... the simplest fix is to [use some horrible hack that disregards the spec]" much more often lately so I'm glad it's not just me.

However I'm not sure how to best prompt against that behavior without influencing it towards swinging the other way and looking for the most intentionally overengineered solutions instead...


My own experience has been that you really just have to be diligent about clearing your cache between tasks, establishing a protocol for research/planning, and for especially complicated implementations reading line-by-line what the system is thinking and interrupting the moment it seems to be going bad.

If it's really far off the mark, revert back to where you originally sent the prompt and try to steer it more, if it's starting to hesitate you can usually correct it without starting over.


That is generically my experience as well. Claude half-assing work or skipping stuff because "takes too much time" is something I've been experiencing since I started using it (May 2025). Forcing it to create and review and implementation plan, and then reviewing the implementation cross-referenced with the plan almost always produces consistent results in my case.

Make sure to use "PRETTY PLEASE" in all caps in your `SOUL.md`. And occasionally remind it that kittens are going to die unless it cooperates. Works wonders.

Can you paste the relevant section in your soul please?

Sure, as soon as I locate my soul.

I love how despite how cold and inhuman LLMs are, we've at least taught them to respect the lives of kittens


Where is that? I found "Return the simplest working solution. No over-engineering." which sounds more like the simplest fix.

I need to add another agent that watches the first, and pulls the plug whenever it detects "Wait, I see the problem now..."

Yeah it’s so frustrating to have to constantly ask for the best solution, not the easiest / quickest / less disruptive.

I have in Claude md that it’s a greenfield project, only present complete holistic solutions not fast patches, etc. but still I have to watch its output.


Time's up and money is tight. The downgrade was bound to happen.

It's a bit insane that they can't figure out a cryptographic way for the delivery of the Claude Code Token, what's the point of going online to validate the OAuth AFTER being issued the code, can't they use signatures?

I think in general we need to be highly critical of anything LLMs tell us.

Claude code shows: OAuth error: timeout of 15000ms exceeded

Maybe a local or intermittent issue? Working for me.

Seems solved now indeed.

That helps explain why my sessions signed themselves out and won't log back in.

I just experienced this some time ago and could not sign in still.

Their status page shows everything is okay.


”I can’t make this api work for my client. I have deleted all the files in the (reference) server source code, and replaced it with a python version”

Repeatedly, too. Had to make the server reference sources read-only as I got tired of having to copy them over repeatedly


Haha yeah. I once asked it to make a field in an API response nullable, and to gracefully handle cases where that might be an issue (it was really easy, I was just lazy and could have done it myself, but I thought it was the perfect task for my AI idiot intern to handle). Sure, it said. Then it was bored of the task and just deleted the field altogether.

Certain phrases invoke an over-response trying to course correct which makes it worse because it's inclined to double down on the wrong path it's already on.

The cope is hard. Just at this point admit that the LLM tech is doomed and sucks.

But it was clearly really food before the regression, the original link (analysis) says as much.

Just because some people try to use a hammer as a screwdriver it doesn't follow that the hammer sucks.

how is it "doomed"?

The cost far outweighs the profits.

i am already on api tokens for the chinese open source models and no subscriptions. these are all available in the original form open source and priced above the inference cost. i think this is the long term option.

zero degradation in speed or quality seen.


So you see better performance with the API plans than the subscriptions?

How complex are we talking? I one shotted a game boy emulator in <6 minutes today

There are countless reference examples online, that's just a slower, buggier, and more expensive git clone.

Yep. If you ask Claude to create a drop-in replacement for an open-source project that passes 100% of the test suite of the project, it will basically plagiarize the project wholesale, even if you changed some of the requirements.

try one shotting something actually original and see how it goes

i keep getting nonsense


A full emulator in 6 minutes? I gotta call BS... I've been working on a SPC700 audio editor in the background as a side project, and implementing the cpu has taken at least 2 hours, and I still haven't implemented all of the opcodes.

This is somewhat in line with the approach taken by some softfloat libraries, e.g. https://bigfloat.org/architecture.html

Related work from our security lab:

Stream of vulnerabilities discovered using security agents (23 so far this year): https://securitylab.github.com/ai-agents/

Taskflow harness to run (on your own terms): https://github.blog/security/how-to-scan-for-vulnerabilities...



I posted this and it picked up steam over night, so I thought I'd add how I'm using it:

I work on 3D/4D math in F#. As part of the testing strategy for algorithms, I've set up a custom agent with an F# script that instruments Roslyn to find FP and FP-in-loop hotspots across the codebase.

The agent then reasons through the implementation and writes core expressions into an FPCore file next to the existing tests, running several passes, refining the pres based on realistic caller input. This logs Herbie's proposed improvements as output FPCore transformations. The agent then reasons through solutions (which is required, Herbie doesn't know algorithm design intent, see e.g. this for a good case study: https://pavpanchekha.com/blog/herbie-rust.html), and once convinced of a gap, creates additional unit tests and property tests (FsCheck/QuickCheck) to prove impact. Then every once in a while I review a batch to see what's next.

Generally there are multiple types of issues that can be flagged:

a) Expression-level imprecision over realistic input ranges: this is Herbie's core strength. Usually this catches "just copied the textbook formula" instance of naive math. Cancellation, Inf/NaN propagation, etc. The fixes are consistently using fma for accumulation, biggest-factor scaling to prevent Inf, hypot use, etc.

b) Ill-conditioned algorithms. Sometimes the text books lie to you, and the algorithms themselves are unfit for purpose, especially in boundary regions. If there are multiple expressions that have a <60% precision and only a 1 to 2% improvement across seeds, it's a good sign the algo is bad - there's no form that adequately performs on target inputs.

c) Round-off, accumulation errors. This is more a consequence of agent reasoning, but often happens after an apparent "100% -> 100%" pass. The agent is able to, via failing tests, identify parts of an algorithm that can benefit from upgrading the context to e.g. double-word arithmetic for additional precision.


> I work on 3D/4D math in F#. As part of the testing strategy for algorithms, I've set up a custom agent with an F# script that instruments Roslyn to find FP and FP-in-loop hotspots across the codebase.

I don't know if there is an equivalent in Roslyn, but in Julia you can have the agent inspect the LLVM output to surface problems in hot loops.


Well sort of, the industry tried to go way beyond that by capturing the entire light field: https://techcrunch.com/2016/04/11/lytro-cinema-is-giving-fil...


Finally, Desert Golf and Flappy Bird merged into one


This guys factory is just across the lake from where I live and this is painful to watch. Both Alibaba and the general local industry (metal fabs, train shops, etc) have high degrees of expertise in supply chain verification. You can hire (heck even bribe) experts along the way to reduce fuck ups. The video contained no mention of any audits, any additional paperwork beyond some pictures.

I once had a company that procured very simple electronics (fingerprint readers) from Taiwan and due diligence included travelling there, meeting every single person in the engineering office in person, then touring the contract factory where this would be built, then negotiating shipping and even driver development details.

This took all of one week and the price of a few plane tickets. We didn’t have the cash for professional auditors. In the end we got a product that worked, and even at a lower price (negotiating at a distance is not effective).


This is what due-diligence looks like. Without it, platform vendors lacking scruples and a proper platform accountability process will cheat because it means more money for them.


This all sounds quite insane. I bet Roman would not buy a car sight unseen from Europe, but here he is sending money in blind to China.


No engine can be 100% perfect of course, the original comment is broadly accurate though. CodeQL builds a full semantic database including types and dataflow from source code, then runs queries against that. QL is fundamentally a logic programming language that is only concerned with the satisfiably of the given constraint.

If dataflow is not provably connected from source to sink, an alert is impossible. If a sanitization step interrupts the flow of potentially tainted data, the alert is similarly discarded.

The end-to-end precision of the detection depends on the queries executed, the models of the libraries used in the code (to e.g., recognize the correct sanitizers), and other parameters. All of this is customizable by users.

All that can be overwhelming though, so we aim to provide sane defaults. On GitHub, you can choose between a "Default" and "Extended" suite. Those are tuned for different levels of potential FN/FP based on the precision of the query and severity of the alert.

Severities are calculated based on the weaknesses the query covers, and the real CVE these have caused in prior disclosed vulnerabilities.

QL-language-focused resources for CodeQL: https://codeql.github.com/


Heyo, I'm the Product Director for detection & remediation engines, including CodeQL.

I would love to hear what kind of local experience you're looking for and where CodeQL isn't working well today.

As a general overview:

The CodeQL CLI is developed as an open-source project and can run CodeQL basically anywhere. The engine is free to use for all open-source projects, and free for all security researchers.

The CLI is available as release downloads, in homebrew, and as part of many deployment frameworks: https://github.com/advanced-security/awesome-codeql?tab=read...

Results are stored in standard formats and can be viewed and processed by any SARIF-compatible tool. We provide tools to run CodeQL against thousands of open-source repos for security research.

The repo linked above points to dozens of other useful projects (both from GitHub and the community around CodeQL).


The vagaries of the dual licensing discourages a lot of teams working on commercial projects from kicking the tires on CodeQL and generally hinders adoption for private projects as well: are there any plans to change the licensing in the future?


Nice, I for one didn't know about this. Thanks a bunch for chiming in!


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search:

HN For You