Yes, this is what I got excited about upon hearing about the recent Adobe "open sourcing."
post
claude aside, i am more interested in starting a project to reverse engineer the binary blobs that infect the linux kernel so it would finally be 100% free
Do you mean the firmware blobs? If so, then be aware that the firmware doesn't run in the Linux kernel, and isn't part of it. They get uploaded to the peripheral devices, and run there. They are often encrypted and signed, so even if you reverse engineer them, and understand what they do, you also need the private keys of the vendor to sign and/or encrypt them before the peripheral hardware accepts them...
This is the repo that people are using, 270k stars
Yea but you will also install other software like maybe gidhra.
I made the fingerprint sensor on my Lenovo 14s Yoga work thanks to Claude. It took 30 minutes.
Please make it available for others to use and not have to spend resources to do it again, ideally upstream it.
I did start a public a GitHub repo right after I made it work.
That's great, could you please share a link to that repository so that anybody interested could benefit from your work?
I have been wanting my fingerprint device working too. Did it require a Claude subscription or would the free plan suffice?
idk about claude but the free version of Kimi is capable of helping you here. I use it for my linux tech support and it's very helpful.
Does Kimi k3 really have a free version?
Well I know the capabilities of the AIs on the tech support side of things, however, free plans have use limits, and often those limits are reached very quickly once you are trying to build such a 'big' project.
I have the pro subscription ($20 a month)
I would love if someone reverse engineered Logitech drivers and made them better as actual drivers instead of an annoying app
Could this one day get us a second life for our echo flex units? They are phenomenal pieces of hardware.
Meta portal TV are also excellent hardware units discarded by its mothership and ripe for reuse. They were so mired by controversy but the TV unit hardware was really great.
You'll be playing doom on your toothbrush in no time.
But yah maybe. There are mountains of "smart" hardware that demanded cloud, cloud went away, suddenly a brick. I'd love to wire some of that up and give Claude a shot at jail breaking and flashing that kinda thing.
Why have all of the LLM apologists come out of the wood work.

If I wanted a virus on my device, I'll just go to a malware repository rather than trust low quality slopware.
Good luck finding a malware repository for undocumented proprietary hardware
Reverse engineering is actually a very good use case for LLMs. Its like using a black box to get a different black box to do a trick. It might not be epistemologically sound, but if it can boot the kernel, who cares?
The great weakness in getting ARM SoCs in Linux mainlined is how long it takes compared to x86. Wouldn’t it be great if AI dealt the deathblow to x86?
Not quite the same, but I recently used it to reverse engineer drivers for a Sony MZ-RH10 Minidisc player.
Built an app that will read and transfer media to the device. Also added transcoding for ATRAC1 to ATRAC3 native formats.
Native tool is on macOS, ported it to Linux for my Arch devices.
As someone who's anti ai this is the usecase I've always kind of advocated for with open source models.
Its conflicting because it still seems to be done by frontier models, and I'm not exactly happy with my monkeys paws wish.
I've always been against the idea of IP rights and thought all things should be a open source platform everyone can work from. Cars, computers, phones whatever. But I never really envisioned a model that consumed everything and destroyed everything else in the process.
I don't like the idea of frontier models profiting off the direct theft of other companies work that is also destroying them. Its much different when you or I slap up a github page with our own open source model on our own hardware, and create a project that often we are consumed by and plan on maintaining it. I don't like a world where now we even don't get options in software, because even software as a service is a dead end. I don't want to have to spend 5 grand and monthly fees to claud just so I can vibe code my own apps because no one makes them anymore. Its either that or I'm sifting through hundreds of thousands of abandoned vibe coded apps with almost identical code, besides whatever "unique" features the vibe coder thought they where adding. It will feel like looking for the one alarm app that isnt shit in the google playstore out of millions.
And whatever we make with it isn't really ours, we are just feeding a machine that needs a exponentially infinite amount of data, who only benifit a microfraction of the population. Just bums me out.
Yeah, kinda sucks that our only option is paying the families of vampires. It would be awesome if the code trained from opensource must be mandatory opensource, i can see local model of LLM being offer by site that host opensource project that trained their model from said opensource, i have quite a few very small personal project i really wanted to do with it but for now i'm holding it off until this AI craze die down a bit. I don't mind it being slow, i don't have the time to do it anyway.
At least this way all those chip won't be horde by big tech.
I don't want to have to spend 5 grand and monthly fees to claud just so I can vibe code my own apps because no one makes them anymore.
I think that's the scary part. When all human maintainers of essential libraries burn out, how will software even be developed? Everything will end up being made by agents, and possibly companies like OpenAI and Anthropic will need to be the new library providers so that agents don't have to reinvent the wheel all the time.
I am tempted to use AI to try and create firmware for RokuTV devices and televisions.
usecase
use case
Just wait until Anthropic makes this against the TOS because it's impacting their friends bottom line.
Can drivers be installed as a Flatpak?
no comments (yet)