Matteo

Software Architect ⚙️ | AI tinkerer 🤖 | Building immersive digital worlds & GMing tabletop adventures with LLMs as my dice 🎲

[Matteo @teomurgi]

Now fixing wifi after wake up MacBook12 2017.
@OmarchyLinux
https://t.co/g7Am2pCf42

[Matteo @teomurgi]

@OmarchyLinux fix for broken audio on macbook12 2017
https://t.co/qS4lpHL3oN
#wecanfixeverything

[Matteo @teomurgi]

https://t.co/mCocYHwh7d <- great optimization for local ai

[Matteo @teomurgi]

@dhh what's the most elegant way to access #omarchy via remote desktop?

[Matteo @teomurgi]

@dhh But nothing in Milan Italy 😢

[Matteo @teomurgi]

@ivanfioravanti have a nice trip! in the meantime you got streaming merged in forged-in-ai 💪🏻

[Matteo @teomurgi]

GLM-5.3 a lot of times gets stuck in extremely long thinking loops full of "Wait maybe...", "Actually wait...", "Or maybe...". So frustrating :(

[OpenAI @OpenAI]

This is GPT-6 Astra.

Anything you can do on a computer, Astra can do for you. Fast.

[Matteo @teomurgi]

@ivanfioravanti @MiaAI_lab @NVIDIAAI ���🏻

[Matteo @teomurgi]

Let's try switching from vscode to @zeddotdev

[Matteo @teomurgi]

@johnternus hi

[Matteo @teomurgi]

To my experience - for coding - Sol >>> Opus 5

[Matteo @teomurgi]

HUD mode of Hermes Desktop is amazing ! @NousResearch

[Vito Tafuni @vitotafuni]

@antirez Visualization is a key concept here! When I approached llm first time I’ve started with an hands-on proj to learn by real examples instead of just theory… https://t.co/iQiXqEo3WZ

[Matteo @teomurgi]

Nice one

[Matteo @teomurgi]

Basically a super-light litellm with headroom embedded.

[Matteo @teomurgi]

Your agent just re-read the same 30k-token log file. Again. And you paid for it. Again.

Headrouter: an OpenAI-compatible gateway that compresses context before it hits your provider.

→ 35.8% on real agent sessions
→ 50MB, no DB

https://t.co/acEPlAJ1Vd

[Matteo @teomurgi]

@dhh Finally managed to install Omarchy as an UTM vm on my M5. steps:
- Create a vm using archboot iso
- Use QEMU -> Apple Virtualization does not allow for 3D GL
- Once arch was installed then proceed with instructions at https://t.co/FAhP9C077x

[Matteo @teomurgi]

Hey Hermes: create a sw development team inspired by bmad method - https://t.co/ng4szxViv8. @NousResearch

[Matteo @teomurgi]

If only there was an agent with the self improvement capabilities of Hermes (by @NousResearch) but the strong tui usability of @opencode