vanillameow

The tagline is literally "No AI" but this is a clearly Claude-designed website and you say you use AI for moderation, so what does the "No AI" even refer to? That the people you talk to aren't AI? What reason do I have to believe this, given the project's remaining track record?

Also IRC still exists.

According to GFN, my ping is somewhere between ~20-25. I can see how it would be playable for the casual audience, but as someone who grew up on native gaming, it's not really playable to me for anything that's not turn based.

It's like how someone can play League of Legends at 80 ping for years, but once you're down to like 10 once you can't really go back.

I really enjoy the prose of this article. The writing breaks down concepts in a very easily understandable way. Thank you for posting, will definitely finish this one later!

I mean it's not as offensive as a lot of other vibe-coded "products", but it's just kind of a waste of everyone's time; there's no ingenuity in the presentation (it suffers from the same emoji-feature-box-small-text combo as every single one-shot vibecoded site in existence) and judging by said presentation I highly doubt the author put the effort into researching this list or writing the comments by themselves either.

So while it's not gonna be the next Moltbook in terms of security breaches, it's basically just the 2026 version of your middle manager copy-pasting a paragraph from ChatGPT web into Slack. It's content from nobody for nobody. It's definitely not what I want to see on the HN front page, but I guess if people get a kick out of it then you do you.

The components heavily give Claude Code vibes. I use CC to build internal tools and, given free reign over the design, this exactly what it will produce.

Won't comment on the writing other than that the punchlines do feel a bit pretentious in an AI kinda way. I've seen the author's blog posts and I much prefer their natural writing to this essay-style output, but to each their own.

Because Opus is kind of degrading lately, I said "fuck it" and made a new OAI account and used the month free trial. I put one query into ChatGPT using 5.5 thinking - the frustrating thing was that it did put more effort into getting correct answers rather than Opus, which is just guessing. Specifically, I asked about the coding harness pi, and despite explicitly referring to it as a harness, Opus 4.7, 4.6 and Sonnet 4.6 all fell back to telling me about Aider or OpenCode and ignored my query completely, while ChatGPT said "I'll assume pi is a harness" and then did in fact find the harness.

However the language of ChatGPT is still the same slop as years ago, so many headings, so many emojis, so many "the important thing nobody mentions". 10 paragraphs of text for what should be a two paragraph response. Even with custom instructions (keep answers short and succinct) and using their settings (less list, less emoji, less fluff) it's still NOTICEABLY worse than Claude on base settings.

I've yet to test Codex, will get to that this weekend, but in terms of research or general Q&A I have no idea how anyone could prefer this to Claude. Unfortunately Claude has seemingly stopped giving a fuck about researching entirely.

I mean, sentiment in this thread (and the neighboring Opus 4.7 one) are overwhelmingly negative this time around. That comment prob would have made more sense around 4.5/4.6.

That said, until models produce verifiably correct work (which is a difficult, if not impossible, bar to clear), I sorta doubt it. Not because humans intrinsically produce better or smarter work (arguably, many humans across many domains already don't vs current models), but because office politics and pushing blame around are a delicate game in corporations.

It's one thing for a product lead to make wild promises and then shift blame to the black box developer team (and vice versa shift blame to the customers when talking to the devs) but once you are the only dude operating the slot machine product generator 5000 the dynamic will noticeably shift, and someone will want someone to be responsible if another DB admin key leaks in production. This sorta diffuses itself when you have 3 layers of organization below you, but again, doesn't really work with a black box code generator.

I think Opus is just an expensive model on API, especially without context management. A single message with near full context (I think this was still on 250k as well) costs like 1$ or something like that iirc.

Imo this is the premium I pay right now to just not have to worry about this. The project where I burned 50$ in a day was using superpowers plugin (A set of skills that makes Claude meticulously plan out design and implementation, interview for details, use subagents for subtasks and review them independently, etc.) - it burns tokens like crazy, but it has super good results for me for custom software tools for myself.

I would probably change my approach if I a) was creating software for customers where I had to actually worry about the implementation details or b) if I was forced to switch to API and couldn't just throw Opus at a 28-task plan for an hour. But this works for me right now so meh. I feel like I'm in some rare Goldilocks zone where Anthropic is not super ripping me off (I use CC quite heavily and generate real value for myself) but I also don't go crazy if I go 2 days without building the next SaaS startup.

Nah I get you, but for me personally, I do use CC a ton. It's building me so many useful internal tools right now, and deep research is also bootstrapping me into some new hobbies etc. I think I'm kind of in a rare-ish (maybe not so much on HN, but for the general population) spot where I'm not really trying to make some SaaS get quick rich scheme, but just directing CC to make apps that would take me a few days to make in a few hours, smoke test them, and solve a problem I had before. (e.g. photo tagging automation, MCP connections for personal services for documentation of chats or setting up todos, ansible playbooks for VMs, setting up data pipelines from external APIs for home automation...)

I deffo get more perceived value out of it than the 100$ I pay. Could I get MORE value with the same 100$? imo only through OpenAI (no harness lock in and more lenient limits), but I deeply dislike the way their company is evolving. Admittedly, recent launches from Anthropic like managed agents and Mythos Preview don't make me very hopeful the individual developer pricing is here to stay, but I'll use what I can get while I can get it.

Could I get my required value with less than 100$? Mayyybe I could get by with like, three Anthropic 20$ plans? or 2x20$ and an OAI 20$? but this is so min-maxy that I just don't really want to bother. Pay by token would kill my workflow instantly. I'd have to add so many steps for model selection alone. I'll cross that bridge when Anthropic cuts me off.

I agree though most people on the $200 plans are either just not using them or in some deep AI psychosis. I'd like to exclude myself from these groups, but the pipeline to AI psychosis seems very wishy washy to begin with (the thread the other day about iTunes charts being AI dominated had a surprising amount of people defending AI music, imo).

I ran this just now and for a small web-app I built I used over $50 in a single day. This was using superpowers plugin and almost exclusively coordinating through Opus. Could I get by with 100$ a month without the subscription? Maybe, but I pay for the convenience of just being able to throw Opus with lavish plugins at it (with 5h limits that are, in my opinion, pretty reasonable). I don't really WANT to have to think about when Haiku or Sonnet are enough.

If anything I would consider switching to OpenAI subscription (if I didn't despise them even more than Anthropic as a company), but converting to API use seems completely infeasible to me. I'd have to severely cut back on my use for not much benefit, other than having maybe an agent thats a little less jank than CC.

1. In reality most people simply do not do this, and frankly it's exhausting to be expected to always assume goodwill in a setting that is full of pure vanity.

2. There's a difference between technical documentation, which AI can be quite decent at, and product marketing. A README is usually about 20/80, maybe 50/50 for large FOSS projects. You can have the AI write the sections on how to install the thing for all I care, but as soon as AI is telling me why I should use it, you've lost me. Signals a complete lack of interest in your own product.

Genuine question - your README is full of em-dashes, emojis, feature squares and ASCII diagrams - none of which are present in your pre-AI era projects.

Why do you expect a potential userbase to care to read something you didn't even care to write?

Seems a bit disrespectful to me.

If you need to author a lot of emails with LLM you should be rethinking your business strategy tbh

I use mainly Opus 4.6.

I did the same thing and created a skill for summarizing a troubleshooting conversation. It works decently, as long as my own input in the troubleshooting is minimal. i.e. dangerously-skip-permissions. As soon as I need to take manual steps or especially if the conversation is in Desktop/Web, it will very quickly degrade and just assume steps I've taken (e.g. if it gave me two options to fix something, and I come back saying it's fixed, it will in the summary just kind of randomly decide a solution). It also generally doesn't consider the previous state of the system (e.g. what was already installed/configured/setup) when writing such a summary, which maybe makes it reusable for me, somewhat, but certainly not for others.

Now you could say, "these are all things you can prompt away", and, I mean, to an extent, probably. But once you're talking about taking something like this online, you're not working with the top 1% proompters. The average claude session is not the diligent little worker bee you'd want it to be. These models are still, at their core, chaos goblins. I think Moltbook showed that quite clearly.

I think having your model consider someone else's "fix" to your problem as a primary source is bad. Period. Maybe it won't be bad in 3 generations when models can distinguish noise and nonsense from useful information, but they really can't right now.

I'm surprised to see this getting so much positive reception. In my experience AI is still really bad with documenting the exact steps it took, much more so when those are dependent on its environment, and once there's a human in the loop at any point you can completely throw the idea out the window. The AI will just hallucinate intermediate steps that you may or may not have taken unless you spell out in exact detail every step you took.

People in general seem super obsessed with AI context, bordering on psychosis. Even setting aside obvious examples like Gas Town or OpenClaw or that tweet I saw the other day of someone putting their agents in scrum meetings (lol?), this is exactly the kind of vague LLM "half-truth" documentation that will cascade into errors down the line. In my experience, AI works best when the ONLY thing it has access to is GROUND TRUTH HUMAN VERIFIED documentation (and a bunch of shell tools obviously).

Nevertheless it'll be interesting to see how this turns out, prompt injection vectors and all. Hope this doesn't have an admin API key in the frontend like Moltbook.