So waking up to hear @opencode was in the midst of a hack sucked.. really felt for @thdxr and the team. To then find out one of my tokens was taken and my google credits drained sucked even more. However I fully support the team and to show it I just upped to a OpenCode Go+ sub

Why am I only just seeing this now

If you don't believe me.. believe @holtbt and hop on that waitlist :)

Ok, idea... phones need local voice models, with an API that we can connect to. Doesnt need to be that technical. Apps can leverage it the same way they do a camera but if i want to manually wire up my own clanker to it.. so be it. @SamsungMobile you can have the idea for free.

this project excites me as it is at the perfect intersection of my increasing love of doing crazy shit with ESP32 devices and being lazy. github.com/Haris16-code/Kryo…

Very few things excite me in AI anymore... Jev from @typesafeai is something that does! Having a lot of fun using it.. will share later.

This is the whole reason we started building

Ok, Fable 5.1 dropped and it finally pushed me over the edge. I’m going to start banging on about this because I’m sick of seeing it, and the new models are making it worse, not better. Had Fable last night tick a box saying a PR met a security check. There was zero record it even checked. Just a green mark and a clean story. No trace. No evidence. Nothing. That’s the whole problem. I don’t trust the success message. I trust what actually landed. If there’s no record the check ran, the check didn’t run. Doesn’t matter how confident the model is. Most teams call this a trust problem. I think it’s just a bad setup. Same model does the work, explains the work, then ticks its own box. Of course it says it passed. If something can write, something else has to be able to say no. And that no can’t come from the thing that did the writing. The frontier isn’t raw intelligence anyway. It’s autonomous thought that can doubt, stop, and refuse. Intelligence without a stop condition is just velocity. That’s the sort of thing I love sitting around discussing with a glass of Scotch. Probably going to start posting a bit about that to be honest... Trust, integrity, agentic systems as they actually behave. Not the demo version people screenshot or have their agent mockup.

I have serious Grok Bot envy.... please @bot bring out a linux, web or android client

Codex creates security issue in code and ejects because is blocked from dealing with cyber security issues...

ahh so someone finally packaged the classic crt.sh recon into something actually readable same CT data we’ve all been using for years… just without the usual mess.. one clean URL and you’re done... nice lol

This project also gives you the ability to add custom items to your acknowledgements which I use to thank the OSS dev tools I use but you aren't required to add to a product advisory.

So @QuinnyPig you are either using X a lot more lately or the algo has finally re-aligned and things are right in the world once more because I am seeing a post from you every time I open the app. Great to see you :)

I just reserved my Cloudflare Wallet tag: aneki.cloudflare.pay. Assuming you give two shits about the upcoming Cloudflare Wallets you can reserve yours now at cloudflare.pay

hey @OpenAI you guys have been getting a fair bit right at the moment and the usage resets are super handy... bit annoying to wake up today on 85% for the week and my reset that was due to expire today... had aleady expired.

Ummmm what??

I have been so buried in agentic development that even though I still use Claude, Perplexity, Gemini, ChatGPT etc for questions, bouncing ideas off and research I totally missed that ChatGPT can now write back to a repo.. and remember a ChatGPT chat can also be linked to 5 repos.