m0rde

It exists by the way :)

https://agents.stackoverflow.com/about

I'm a bit confused. You posted fake stats online and saw ChatGPT visit the page. So you immediately assume that ChatGPT is falling for those fake stats and using it in chat responses or analysis input? Where's the proof for that? How do you know ChatGPT didn't search, find and discard the fake stats?

You also saw your fake stats in a paper and a few other sites and assumed ChatGPT put it there? Not a human visiting your site and grabbing those fake stats?

Did I miss the evidence that shows ChatGPT is actually using the fake stats? Serious question.

P.S.: Posted this as a top level comment then saw your post.

I'm a bit confused. Someone posted fake stats online and saw ChatGPT visit their page. Then they immediately assume that ChatGPT is falling for those fake stats and using it in chat responses or analysis input? Where's the proof? How do you know it didn't search, find, and discard?

The author also sees their stats in a paper and a few other sites and assumes ChatGPT put it there? Not a human visiting the fake stats site?

Did I miss the evidence that shows ChatGPT is actually using the fake stats? Serious question.

You're rude

Thanks, great ideas to try out here!

Enterprise pay as you go across a few vendors. But still cost conscious.

Defaulting to Sol Medium/Light for most planning and implementation I think is tricky or want more care in.

Luna Extra High for everything else (implementation, tedious take over my browser and do stuff).

I read all of its output tokens and lots of thinking tokens to understand the general flow of things, but only minimally look at code these days. I can't grok what Claude models speak and it's gotten worse. OAI models speak my kind of tech language I guess.

Light human review, some automated review.

Most work is for internal use.

Do you just go around posting this comment? <https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...>

> Fable 5: Reasoned so long it exhausted the output budget before finishing the drawing.

Lol

> * Grok 4.5 has an advantage on CursorBench: an earlier snapshot of the Cursor codebase was unintentionally included in training. The exact score impact is unclear. That data has been removed for future models. For a rundown of third-party benchmark scores, see the Grok 4.5 launch blog.

I don't know about those numbers, even assuming this was by mistake :)

Mighty Squirrel is at Gore place for the summer...

Okay, maybe that's beer and closer to Waltham =P

Yeah. I get that many HN comments are just complaints (heck mine was too and just as negative and shaming). But how bad of a day must you be having to try to shame someone about how they choose to write up an experience they thought was neat. Whatever, free speech and all that. Hope OC's day gets better.

What a difficult world you must live in these days

From the post:

> If you care about how your content moves through the world now, including through AI systems, you have to care about caching. Not as a performance optimisation for human browsers, but as infrastructure for machine readership.

> If a comment just mentions Opus without being more specific and in the absence of relevant context clues, it gets mapped to Opus Latest

Consider keeping this data point but instead calling it something like "Opus Unspecified". Let the user decide how to interpret it.

New technology isn't perfect now -> drop technology and never use it in the future

I think about what I do in these verbose situations; I learn to ignore most of the output and only take forward the important piece. That may be a success message or error. I've removed most of the output from my context window / memory.

I see some good research being done on how to allow LLMs to manage their own context. Most importantly, to remove things from their context but still allow subsequent search/retrieval.

Feedback loops are important to agents. In the article, the agent runs this build command and notices an error. With that feedback loop, it can iterate a solution without requiring human intervention. But the fact that the build command pollutes the context in this case is a double-edge sword.

July 2025, then posts a few days later :) maybe he just meant done with the startup life and kept blogging. Hope you find that stability.

> The noticeable spike [~20 percentage points] in May in the figure above [tool invocations] was largely attributable to one sizable account whose activity briefly lifted overall volumes.

The fact that one account can have such a noticeable effect on token usage is kind of insane. And also raises the question of how much token usage is coming from just one or five or ten sizeable accounts.

Good points here, particularly the ends not justifying the means.

I'm curious for more thoughts on "will drive more and more people out of jobs”. Isn't this the same for most advances in technology (e.g., steam engine, computers s, automated toll plazas, etc.). In some ways, it's motivation for making progress; you get rid of mundane jobs. The dream is that you free those people to do something more meaningful, but I'm not going to be that blindly optimistic :) still, I feel like "it's going to take jobs" is the weakest of arguments here.

Many people would not have found this cleaner without Googling, reading reviews, etc. While that may not be an ad directly, it's part of the marketing budget. So what needs to change?

This is cool. I'm interested in reading more on this concept. Any chance you have a write up of experiences so far? Or can you point to other resources?

This looks fun, thanks for sharing. Will definitely give it a shot soon.

I read over the repo docs and was amazed at how clean and thorough it all looks. Can you share your development story for this project? How long did it take you to get here? How much did you lean on AI agents to write this?

Also, any plans for monetization? Are you taking donations? :)

Have you tried giving it tests? Curious if you found they made things worse.

Really well written README :)

I've never used Claude Code or other CLI-based agents. I use Cursor a lot to pair program, letting the AI do the majority of the work but actively guiding.

How do you keep tabs on multiple agents doing multiple things in a codebase? Is the end deliverable there a bunch of MRs to review later? Or is it a more YOLO approach of trusting the agents to write the code and deploy with no human in the loop?

> Customer support lines/banks/etc must shift their opening hours slightly so that they are open outside of business hours, so that working people can use their services

How would customer line/bank/etc workers ever get help with these services?

Is there an example we can see?

Original post: https://x.com/sama/status/1899535387435086115

Great idea, congrats. Can you speak a bit about the the validation piece? Were LLM hallucinations an issue and required this? Are you using some kind of structured output feature?