Spent the last few days vibe-coding this. https://dstretch.com/Apps/index.html isn't real time and is $20. Worse, if you shove it a non-RAW image, (at least in my quick try) it just highlights the JPEG compression. I couldn't find anything that did it real time. Threw some tokens at the problem, made it use RAW sensor info.
Edit: Only once? Maybe an ad is taking over the page sometimes? Something is definitely weird.
BaudGirl's VA is probably the biggest right now: https://baudgirl.com/work/FM-1+VA
But there's also https://github.com/hugelton/Felucca, https://github.com/charlesvestal/fm1-x0x, https://www.groove-os.com, https://isod89.github.io/sloop-fm1/webapp/installer/, and https://github.com/charlesvestal/fm1-fomni
AFAIK, all of them are vibecoded, but that doesn't mean they're not super fun.
Edit: Someone made a good firmware list: https://fm1-editor.com/firmware/
Lots of new tools have been popping up lately ranging from made by hand to made by vibes. Some of it good, some of it bad. What gems have you found that are from the last year or two?
"Self healing" doesn't sound like a feature to me. It sounds like the product so regularly borks itself it needed nasty safegaurds put in to not, and it was done in a weird way that's annoying to diagnose and fix. Your most stand out feature being it doesn't break it self isn't great. Otherwise, it's what? Fancy on-prem Alexa? Why wouldn't I just setup Home Assistant Voice?
I suspect normal HN readers aren't your target market, but I don't really see who is?
and
> Models will become so fast that UI will be generated on the fly
Tells me this author is a bit, uh, disconnected from the real world. Your time to UI being populated and the UI's familiarity are the product.
I'm working on a huge vibe coded app in Rust rn, but it's entire purpose in existence is to be way, way faster to open and use than the software it competes with. The difference in feeling of something that opens and is usable within ~10ms (so under 1 frame at 60Hz) compared to one that's 100ms or 1s is HUGE. Getting it to be fast, good UI isn't a single prompt, it's a huge amount of refinement, over many weeks.
The difference between software with very straight forward keyboard shortcuts and intuitive UI, guided and crafted for a human, is huge.
Right now, models particularly struggle to make good UI, because it's a human problem. It doesn't care how big text is, because it's reading HTML, not visually looking at the page. Even if you feedback loop it image processing, it's a long ways from making good UI, especially good UI 1-shot. Even if you assume this makes leaps and bounds AND that "frameworks" that make the UI share common design language across on-the-fly prompts AND that it caches UI's that the user likes, etc. etc. I think this is incredibly far off base.
Then there's the missing points of software not being all basic tools that the end goal is to make a SAAS for. Embeded software exists. We're a long way from it being smart enough to figure out when something is a hardware issue. Hardware harness that connect it every test point are ungodly expensive, and so will continue to need a human in the loop for the foreseeable future.
Or, to pick on different points
> Code review will die. [...] Unit tests might die too
Alright, but then you're first up to use the AI vibe-coded pacemaker, or large machinery, or medical diagnostic hardware, or car.
To quote Doctorow,
> There are no cars, only computers we sit in. There are no hearing aids, only computers we put in our ears. There are no 3D printers, only computers that drive peripherals.
I do think AI will change software development, but not everything, not universally, and not without consequence to quality that we'll have to learn the hard way.
I tried to get my lawyer-mom switched to Linux a bit ago, and she loved it in generally, but none of the Office competitors had good enough compat to work.
The only other thing keeping her on Windows is Adobe PDF pro, since it can do OCR where, when you edit it, it reflow the text in a font that matches the scanned in one to look like the original. (I also got weird "This feels like it enables fraud vibes" from this, but, no, turns out it's a totally common workflow for lawyers to need to do this ... I hate it.)
Cool to see 1/2 of the problems keeping her on Windows solved.
I also pretty much exclusively work at my desktop - if you use multiple systems I can see where this matters.
I have some custom skills/MCP servers I use:
one MCP server for using https://usefeyn.com/blog/pulpie-pareto-optimal-models-for-cl... ran locally so that reading web pages is cheaper and uses up less context
One MCP server for doing automated code quality checking, using Valknut, Jscpd, and Lizard to output scores, combine/aggregate them. That's then give to the model so that it can see if there's a mess of code dupe or if there was significant architectural regression. It's allowed to say "This is worth it" but it ground it a bit and stops really stupid, short sighted, hacked-in features
Another uses https://github.com/thejens/id-token-nicer so that the model doesn't bork a big ID without being noticed. It's told to translate in and out of that as necessary.
I had it write it's own skill for making sure to follow some sane commenting/style guides per language, that way it doesn't just give me massive files with no comments.
Finally, a last one lets Antigravity or Codex call out to local models as workers, for long running background tasks where a cheap, dumb model is fine. Saves on overall token usage.
I also have a few skills from the internet setup. https://github.com/AlmogBaku/debug-skill being the most important one.
On top of that, the way people work is notable: Do you use worktrees for agents? Why/why not? Do you have only one subscription? What level? I'm using both Codex at $20/mo and Antigravity at (promo) $5/mo right now. Antigravity's TOS says you MUST use their harness, which informs some of my skill setup/workflow. Do you make your agents.md self modifying, write it yourself, or not use one at all? Do you use a sandbox or YOLO it?
Even just how you prompt matters. On Antigravity/Gemini, I can give it a big list-o-TODOs and have it make a very detailed implementation plan itself, carry it out, and generally do the thing, all unsupervised. Codex, from what I have experienced, isn't as good at that. Doing the planning as a separate step, writing to a file, then telling it to step-by-step it with commits for each step makes it work okay again though.
I'm a reasonably heavy AI user, and have some custom skills/MCP servers I'd share, but there is no way in hell I'm connecting to some arbitrary MCP server and connecting my Github account to it. Noppppeeeee.
> Continuity (cnt for short) is the git storage system [...]
I guess someone realized that turned this into even more of a joke than it already is.
[1]: https://web.archive.org/web/20260818210312/https://cursor.co...
Q: Can I run Proxmox VE on a Raspberry Pi or a similar single-board computer? A: No. An arm64 host must boot through UEFI and describe its hardware through ACPI. Device-tree-only boards and legacy boot are not supported.
Basically no consumer facing arm SBCs, NAS boxes, etc will work with this. Frankly, I see this as a good thing though: Maybe it'll make the SBC market get its stuff together and start actually doing UEFI instead of every board needing a custom build.
If you use Cardnial as a VST you can use any of hundreds of generative sequencers it has.
Browsing midi clips compared to something with parameters that you can tweak, like all of these have seems infinitely preferably than wandering through a sea of MIDI clips.
There's a million better generative MIDI tools out there, and the vast vast majority are free. I can't fathom making this and not being embarrassed, let alone anyone actually spending money on the site.
I would normally reserve such harsh criticism but as sporkl points out, this seems to be some LLM generated drive-by money-grab mess.
I do worry that maybe OP is like a teenager and I'm coming off overly harsh, but even if that's the case, then OP needs to learn what actually makes a product differentiate itself and how to actually appeal to a target audience. This ain't it chief, as a musician, I find it actively offensive.
If you think about all the projects you don't have time to make that require code but would be really cool albeit have no *marketable* value, making those faster to make and easy to share isn't a bad thing.
I want more cool free things people make out of passion - sure, you could argue using AI removes some of that passion, but there's also a large subset of people who are passionate about their field but not passionate about code, and if they're able to make something cool by feeding the idea in and pulling the token generation slot machine's lever on repeat to get their vision, I still think that's cool.
Of course, there's a line where it's slop, so it depends what they're making. A tool to make music? Cool. An album where it's all AI gen'd audio. Not cool. A tool to modify art/apply filters/modify brushes? Cool. AI art standalone? Not cool.
Basically, is the target something standalone as a product we want to have human creativity in the output expression (art) or not. I don't think of MS office as particular artful, but I'm sure many good books have been written in it.
This line is definitely blurry and full of gray areas. For example, https://www.redwoodrhetorica.com to me is totally fine, but I could see why people find it weird.
Similarly, I'm sure to someone working in or on emacs or vim, they're almost sacred and they view the tool itself as a work of art, such that the idea of using AI to improve either sounds offensive, but as long as VSCode works (which, it has had more bugs lately...) I really don't care if they used Claude or whatever to work on the editor/IDE itself.
Of course, there are projects and features which probably shouldn't make it past the "Should this exist?" filter. Complexity does have a cost - nobody wanted CoPilot in Notepad - but having LLMs doesn't change that, I don't think. It means we can do more, but being selective and having good taste to avoid making something bad by adding unnecessary crap to it was a problem far before LLMs.