GitHub에 올라온 에이전트 스킬 상당수는 효과가 검증된 적이 없고, 실제로 별 도움이 안 될 수 있다. 올해 나온 연구들이 이를 보여 준다. 공개 SWE 스킬 49개를 실제 저장소 과제에서 시험했더니 39개는 통과율을 전혀 올리지 못했고, 평균 향상은 +1.2%, 3개는 오히려 성능을 떨어뜨렸다(SWE-Skills-Bench). 공개 스킬 3만 4천 개 중에서 에이전트가 직접 골라 쓰게 하면, 조건이 현실에 가까워질수록 이득이 줄어 스킬이 없을 때와 비슷해진다(Liu 외). 토큰사용량 통계를 보면 스킬을 썼을때 400%가 넘는다. 그러나 잘 선발된 스킬의 통과율은 16%p 오른다. 스킬자체가 문제가 아니다. 검증 안된 스킬이 문제다. 심지어 네거티브 퍼포먼스까지 가능하다. 아무것도 안하는 것보다 못한 선택이 될 수 있다는 거다. 근데 이 스킬검증이라는 건, llm의 stochasticity 땜에 인간이 하기 어렵다. 그러므로...
Taeho Kim retweeted
We've been early adopters of Rust since 2018 and a lot of our infra, voice tech, and even some of the current client is already Rust, the holy grail was getting the whole client there. That means Electron goes away, and so does React Native, mobile keeps native UI but shares app code with desktop via Rust. We hope to get it into testing with users soon™. (And it's not Tauri, it's GPUI, so good guess everyone.)
Alright, back to staring at the WoW Forever login screen.
Taeho Kim retweeted
I've been reviewing a lot of code written by Opus 5.5, and it still seems to make a lot of bad software engineering decisions. The kind that will accumulate/calcify into unmaintainable software over time. IMO even the agents will struggle to maintain vibecoded software.
Taeho Kim retweeted
Cloudflare is taking ownership of tsrs, a port of TypeScript to Rust. It's 8x faster and uses 70% less memory than TypeScript 7.0.
We are not forking TypeScript – we are working with the TypeScript team and as a result 7.1 is already ~2.5x faster, too.
blog.cloudflare.com/typescri…
Taeho Kim retweeted
대기업이 한순간에 무너졌어요 🤯
한 미친 오픈소스 개발자가 AI를 사용해 뭐시기 소프트웨어를 100% 복제했어요
- 구매 필요 없음
- 구독 없음
이제 뭐시기 기업은 흙만 먹고 살아야 해요
(아직 만들다만 오픈소스 링크)
이걸 그만보고싶음
Taeho Kim retweeted
"Almost every software development organization has at least one developer who takes tactical programming to the extreme: a tactical tornado. The tactical tornado is a prolific programmer who pumps out code far faster than others but works in a totally tactical fashion. [...] In some organizations, management treats tactical tornadoes as heroes.
"However, tactical tornadoes leave behind a wake of destruction. They are rarely considered heroes by the engineers who must work with their code in the future. Typically, other engineers must clean up the messes left behind by the tactical tornado, which makes it appear that those engineers (who are the real heroes) are making slower progress."
@JohnOusterhout, A Philosophy Of Software Design, 2018
Sound familiar?
Taeho Kim retweeted
Imagine the amount of Java crashes this poor LLM must have in its training set that it decided 'NullPointerException' deserves its own token
Taeho Kim retweeted
👀 Introduce pulls.review
Categorize changes in Pull Requests into logical groups, with a summary. Helps you do reviews more efficiently by focusing on what matters.
With opt-in GitHub userscript, GitHub Actions, or CLI for local git changes!
Taeho Kim retweeted
A framework is an intelligence cache.
You could waste tokens reinventing the ORM, cache, queues, auth, routing, etc, or you could just pick a world-class framework.
>Anthropic said nearly a quarter of its revenue came from two customers last year
who are they?
Anthropic IPO prospectus leaked to Reuters
Anthropic reported a net loss of $42 billion in 2025, and plans to spend $518 billion on cloud, computing and infrastructure obligations in coming year, according to the prospectus.
The prospectus details how the company has grown sharply in the last year — while also posting wider losses. Revenue grew 12-fold in 2025 to nearly $4.6 billion, even as the company lost more than $8 billion on an operating basis, excluding writedowns of various liabilities mostly tied to previous fundraising, according to the documents, reported here for the first time.
reuters.com/business/finance…
Taeho Kim retweeted
Reject non-understanding.
Slop grenades are not just code and PRs. We run a real risk that *reading* gets entirely discounted, because of the exhaustion of getting low-quality and unverified AI prose being thrown at you all the time.
There's a thread going viral about a 'perf improvement' attributed to a compiler change. The PR *description* itself has the AI telling this person it's *not* due to the compiler change, it's how the algorithms and data structures have changed. And the implication is that a similar win could have been attained otherwise.
I want AI in the service of understanding the universe and enhancing human cognition and creativity.
Taeho Kim retweeted
コーディングで「AIオーケストレーションッ! ループループッ!!」とかやってみた時期あったけど、こういうの何度か踏んでからは指揮者絶対ゆずらないマンになった
レビュー「[P2] 10円玉50枚でコーヒーを買うケースが想定されていません」
AI「指摘は妥当です。50枚同時に入れられる10円専用投入口を用意します」
レビュー「[P3] 1円玉500枚で買うユーザーがいる可能性もあります」
AI「指摘は妥当です。コイン投入口を10倍に広げます」
ワイ「どれどれ………何コレ?」
Taeho Kim retweeted
The more time I spend working with coding agents, the more convinced I am that they make software engineering even harder
We can do amazing things with them, but unlocking their full potential requires extraordinary discipline and knowledge
Replying to @GergelyOrosz
Bury your head in the ground at your own risk. I aim to not jump on any hype trains, but since Opus 4.6 and GPT-5.2 + the harnesses it was clear that these things can write code nearly as good as I can in my best language; better in other languages.
But we won't see non-devs push production code for a long time; possibly forever, if you ask me. Software still needs to be built in robust ways, and it's our profession to do this well, and use the new tools we have, which create new and interesting (+tough!) challenges
Taeho Kim retweeted
I have absolutely never ever "asked" to stop R&D on LLMs.
In fact, I was very supportive of LLM efforts at Meta. Just look up my public advocacy for Galactica (November 2022), and the open sourcing of OPT-175B (May 2022), and of the Llama herd (mid 2023).
I was actually influential in convincing the Meta leadership to start an AI-first product division built around Llama in January 2023.
I've always said that LLMs were impressive, useful, and had amazing potential.
I've also always said they were an off-ramp on the path towards human-level AI.
Stop making me say things I didn't say or claim that I somehow hindered LLM R&D at Meta.
That's BS.
Taeho Kim retweeted
"DiffusionGemma as Jev" showcases the power of non-autoregressive architectures.
While Jev demonstrates the value of rapid decision models, running DiffusionGemma in this paradigm leverages canvas diffusion to evaluate structured choices in a single parallel pass:
⚡ ️Massive Parallelism: Denoises across an open canvas in a single step instead of sequential autoregressive token generation (~0.2s on a DGX spark).
🧠 Full Bidirectional Attention: Allows every option to attend to the full context concurrently, yielding well-calibrated decision distributions.
👁️ Multimodal Grounding: Inherits Gemma 4's spatial vision capabilities for complex visual and text decisions.
Read more about this approach here:
github.com/vllm-project/vllm…
xitter.cf/mmastrac/status/210037…
xitter.cf/mmastrac/status/210062…


































