Redesigning my IDE for agent-assisted workflows

i had a lot of fun chatting with @mattpocockuk today about how i was able to land 2,500 PRs last month! Matt is a wonderful interviewer so i think the interview turned out really interesting both of our skill plugins work great together, so i recommend giving both a try and picking the best skills that suit your workflow piped.video/watch?v=MN9dGgmL…

That's what you get for updating Omarchy. Year of desktop Linux? Not quite yet. Keep your files backed up, folks.

I'm making my third attempt at Pi.

there are only 2 things I want from LLMs: 0) no unnecessary code 1) put the code into right places 2) no bugs how hard is that?

openai announces glm-5.3 in codex subscriptions anthropic announces glm-5.3 is evil and chinese

Made the agent report costs of each operation. Very interesting. OpenCode, Sol 6.1.

What will happen if a spaceship shoots an incoming asteroid with a laser? It will get hit by a molten asteroid. /shower-thoughts

I am going to post a new article about review skills next week. Any questions/topics/ideas? Only hand-typed letters, no AI! (I'll only fix the grammar because I am still bad at it.)

NADIR: IT DOES YOU (final version) My response to the Astra vid

JetBrains will completely lose IDE business if they won't add some crazy AI features like interactive voice code editing. I want to look at the diff, talk to the app and see the code changed in real time. Also switch between worktrees without starting 5 instances of the IDE.

9 games I had the most fun playing

WIP but ~10.7x faster than the original Rust code and 1.25x faster than the asm version on average. Still written in Rust with AVX-512/AVX2/SSE2/Neon opts github.com/omacom/ttfx/pull/…

The current peak of progress lies in refactoring and optimization skills. Yes, LLMs produce slop. But it can be fixed with LLMs. Every time we improve our review skills, we improve the entire codebase, or even multiple projects. I even made a skill that improves review skills.

Oh come on. Quick look at colorshift (13x faster): HashMap in hot loop (unoptimized Rust vibeslop) is slower than indexed arrays (ASM). No shit. Numbers like 17x should raise eyebrows but reading the comments nobody is thinking anymore. This AI psychosis has fried people's brains

I would believe in up to 2x speed increase max. But realistically, knowing that DHH does not read Rust, there is an obvious explanation to 17x. Rust implementation was a terrible vibe-slop from the beginning, and even replacing it with another slop improved the performance.

The first app I ever sold was only 5 bytes. (Not a typo.) I was 16 years old. Now, "Hello World" takes 10 million times more. How time flies...

LLM language: if (condition that is always true) do the thing fallback: code that breaks your app if it runs

There are many states between works and doesn't. For example "occasionally works", "barely works", "works on my machine", "some tests pass", "bug monitor spikes on Wednesdays".

"I love Rust! Rust is amazing ...if you never, ever, EVER have to look at it yourself." — @dhh

Slop is an error-prone implementation. Yes, it is up to the reviewer to decide, but "will we have a crash in production" can be realistically checked via Crashlytics and similar tools, so it is objective enough. Come back to earth, we have grass.