Daniele (@ildella) | teapawt

Tweets
All Posts
Sort by
Most recent Popular
Tweets & Replies
Reposts
Media
Search

Daniele retweeted

Nucube@nucubemusicOct 22

Nucube 0.12 has been released on September 17: our first open release. Music on an SMB or NFS share plays straight from the share, streamed as you listen. The same app on desktop and mobile.

1

3

Daniele retweeted

Nucube@nucubemusicOct 21

Nucube is a music player for the files you already own. It reads local folders, NAS shares and DLNA servers, and leaves everything where it is. The same library on desktop and mobile. No Nucube server, no box to buy.

2

1

26

Daniele retweeted

@deepseek_aiOct 17

🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4

1,622

37,158

8,055,681

Daniele retweeted

@callebtcOct 14

Shoutout to the creators of Kimi K3 @Kimi_Moonshot for providing a ton of inference to the Bitcoin Red Team. Moonshot is the only lab that has stepped up to help the Bitcoin community. Let's make open-source software more secure together!

6

734

15,490

I mean... you have people working at major companies like this that even accept a comment in a PR? Jesus Lord save us all.

2

14

This is true. I mean, this generic statement per se is meaningless but what he means behind that, is true. At the same time, it is also true that one cannot disregard the quality of a codebase. Or one cannot build anything larger than a prototype. It's not "muh, assembly", shut up. People spent lifetimes optimizing assembly. Not happened yet for our shit.

so many of you claim to be obsessed with code quality and say you're capable of producing incredible stuff i'd never believe you should be going crazy with agents right now. refactor everything. everyone is incompetent except you, rewrite their code you're so good please we need you

33

LOL. Can you imagine the simps "open source community devs" building Linux stuff and saying they do not accept AI contributions? I mean... Fuck. I believed I was over accepting most people are not able to accept change. But damn.

@linuxiacOct 12

Linus Torvalds experiments with AI-generated Linux kernel code, using Claude Code to optimize memory operations and improve read performance. linuxiac.com/linus-torvalds-… #Linux #Kernel #OpenSource #AI

Linus Torvalds Experiments with AI-Generated Kernel Code, Sees 3x Speedup

Linus Torvalds experiments with AI-generated Linux kernel code, using Claude Code to optimize memory operations and improve read performance.

18

For the nTh time this year, agents were shipping shit. Why? Cause the codebase of the semi new project was not at the proper level and they've been disabling rules inline. I said it before and I repeat it:, a sane codebase is number one priority (also) in this time and age.

19

Now SayIt for Linux is waiting for @OmarchyLinux publications: thumbs up on the PR: github.com/omacom/omarchy-pk… 🙏

Add sayit-bin, the Say It local text-to-speech app, to the fast ring by ildella · Pull Request #813...

Summary Add sayit-bin, the Say It local text-to-speech app, to the fast ring. Say It turns copied text into speech with open models (Kokoro-82M via kokoro-js / onnxruntime-node) running entirely on...

@callebtcSep 16

another sayit port for linux appeared!

3

1,000

As I am building LACE - my "agent automation" - on top of board -> harness -> GH PRs, I am also still using GrokBot (which I use since day one). I'm almost not using it for coding, but for *everything* else. I have the chief of staff, couple project coordinators, social, legal, product engineering, master of computers. It's everywhere (Desktop, Mobile), is connected to apps, can control computer tasks (and LACE!). @bot

1

41

I'm barely using the TUI anymore. I just use it when on planning sessions they ask questions.

Replying to @ildella

Now it assign a linear card to opencode session. Then lace advance will assign the review when the coding session is done. And then a rework or request to merge. And now I'm building full automation: "lace go" is running moving forward all my tracks. "lace board" is becoing reactive and showing a unified view of linear card, coding sessions and pr/ci status... And I have automation for many automatic "advance".

1

14

Daniele retweeted

SpaceX Super Intelligence

We are glad to support the Omarchy team

Thrilled to welcome @spacexai as a Founding Corporate Patron for the Omacom Foundation. That's $1,500,000 in @grok tokens for the maintenance and development of Omarchy. Maybe soon they'll be beamed straight from orbit? Let's go 🚀🌕 omarchy.org/news/2026/10/spa…

SpaceXSI joins as a Founding Corporate Patron with $1.5 million in Grok tokens - Omarchy News

SpaceXSI is joining the Omacom Foundation as a Founding Corporate Patron, donating $1.5 million in Grok tokens! That’s a serious stack of tokens to further accelerate the agents and teams that build,...

142

6,633

256,979

I've stopped other developments to complete LACE - my personal software factory. It now automates the development workflow: code/review/rework/approve. Now the track is: linear card -> opencode sessions -> githug PR. I'm building LACE using itself. It's doing good. ...continue

1

1

48

The whole thing has been built by itself. I did set the tone of the code with the first CLI at the beginning of this 4th iteration but in this last rush I'm not ever watching a line of code. I have setup a massive amount of code check and rules from lint to duplications etc....

1

9

On top of that GrokBot is my team lead: we define new specs, check everything is working. It can be the user of LACE as well. Now it's having a look to see if there's some common pattern in a potential problem: we have too many review asking for many reworks on same PR. There must be something wrong... The End (for now)

7

With Opencode Go, the monthly cycle also resets the weekly usage mid-flight. That's classy.

2

35

Daniele retweeted

This is particularly true in the world of SaaS! We sucked up so much talent to grind out infinitesimally improvements in, say, expense tracking that could have moved the needle on more ambitious missions.

17

159

37,411

In fact, there has never been a moment in which the quality of code in all it's aspects has been more important. The greatest improvement in agentic-development performance I had this year has been every time I fixed the structure, added even more rules and checks. Oh and about the "but unit test useless", I have a news: unit tests have (almost) always been useless for apps. It's nothing new.

Replying to @dhh

Nahh... The Loc count today as much as yesterdays.

1

1

22

What is also true today is that it's possible to build and ship relatively small projects that deliver great value fast, without even looking at the code once. Prototyping and duplicating one-shot is now massively affordable.

1

12

Will everything be like a quick fast prototype? Maybe. Not today. This is still the same game played with different tools at different speed. The increment in speed changes some rules not the whole game. There will he different styles and approaches like always.

3