Has anybody solved the computer use problem on Linux? I want computer use on Omarchy that is as good if not better than Codex's on Mac.
I saw @trycua 's announcement but I have yet to find a really good plug-and-play computer use tool for agents running on Omarchy.
@hunvreus I disagree.
Obtaining knowledge is never a bad thing. There's a reason that the best builders using AI today come from a background that forced them to learn the old fashioned way.
To gain enough knowledge to become great, you have to start somewhere.
@VictorTaelin This is awesome!
What does it do to caching? Does everything get messed up because you're rewriting the chat history on every request? Does this make it significantly more expensive?
@brandon_galang You're both correct.
Learn to drive at the highest abstraction level (good phrasing btw)
AND learn to code the old fashioned way.
You don't know what you don't know. There's a reason that smart people build good things. Knowing how things work is important.
@rmoff The problem is that it basically makes people skeptical of everything. Build something by hand and people will still question if it was built by AI.
I'm curious how we get out of this loop of everyone generating stuff with LLMs. Feels like it's basically required to move fast.
@KeithZhai@kunchenguid I’m building one rn and it’s going fairly well. I’m using the lessons that Kun teaches. I’ll make this public if I think it’s worth sharing.
First day using Omarchy was a blessed experience thank you @dhh for building such a great product. I’m excited to never use a windows machine ever again.
@kunchenguid Kun could I ask some advice? I currently use the $20 ChatGPT plan and I'm looking for an upgrade. What is the best "cost per intelligence on a given task" plan that I could upgrade to? I'm less familiar with the grok subs so idk if those are in the running.
I see all of these people talking about running agents on other machines. How many people ACTUALLY have another persistent machine to run agents on?
I feel like I'm the only one missing out...
I think Mitchell just gave me another reason to want to code by hand.
Nobody can create excellent projects like this with AI unless they truly have expertise. You cannot gain expertise without struggle and hard work.
For me, that looks like coding without AI (trad coding).
Quick Superlogical demo to end the week. This demo is just of the basic functionality, but I want to highlight just how FAST it is. There's a WHOLE lot more functionality already present and coming but I wanted to keep the demo short. But seriously, check out that speed.
I'm showing the macOS app here and I get asked a lot so let me just say up front: its not macOS only. We're still figuring out exactly what platforms will be stable enough for the initial public release, but we're working on others. In particular, the web interface is very functional but I'll demo that another time (or someone else will).
This is also all showing off the architecture I previously talked about for terminal multiplexing which is significantly different from other mainstream multiplexers: this is all a custom binary protocol where the server is maintaining N replica distributed terminal state machines. It works!
Tons of hard work going into this by the team @almonk@dizzyup@pearkes. Hope we can bring you something you can use soon. ❤️ Happy to answer questions in the replies.
I'm kinda annoyed by how good the unslop skill is. Why has this one simple skill made it so much more enjoyable to work with agents? @poteto you are the goat.
@kunchenguid@ssbrouhard@poteto unslop skill unironically made me enjoy coding with AI again.
I feel like an idiot for not realizing sooner that AI responses were so bad...
@RhysSullivan I fundamentally disagree. Have you seen AXI?
https://t.co/tZIa8TEuO2
MCP may not be bad in principle, but most MCP implementations are not optimized in a way that can achieve the same performance of its AXI CLI counterpart.