よこさん

@dansyu_callenge チャットでしたか、すみません！！！Codeと勘違いしておりました( ;∀;)お返事ありがとうございます！

@yokotin34 すみません！ コメント見落としてました🙇 よこさんも指揮官つくってらっしゃるんですね✨ チャットの指揮官、勝手に動きますか？僕のはそう言うことないですねぇ。はやくチャットのプロジェクトをcoworkやCodeで使えるようになってほしいですね！

GitHubで22,000人がブックマークしてる Claude Codeの教科書があります。 でも全部エンジニア向け。 「Claudeのプロジェクトで 設計AIを作ってClaude Codeに渡す」 なんて一言も書いてません。 非エンジニアの運用法は、 まだ誰も体系化してなかった。 だから書きました⤵︎

サーバー買っちゃった!!

@alex_prompter In fact, reasoning ability and hallucinations are not fundamentally related. The higher one’s reasoning ability, the greater the clarity of one’s dreams.

🚨 BREAKING: CLAUDE JUST GOT NERFED. AMD’s AI director just analyzed 6,852 Claude Code sessions, 234,760 tool calls, and 17,871 thinking blocks. Her conclusion: “Claude cannot be trusted to perform complex engineering tasks.” Thinking depth dropped 67%. Code reads before edits fell from 6.6 to 2.0. The model started editing files it hadn’t even read. Stop-hook violations went from zero to 10 per day. Anthropic admitted they silently changed the default effort level from “high” to “medium” and introduced “adaptive thinking” that lets the model decide how much to reason. No announcement. No warning. When users shared transcripts, Anthropic’s own engineer confirmed the model was allocating ZERO thinking tokens on some turns. The turns with zero reasoning? Those were the ones hallucinating. AMD’s team has already switched to another provider. But here’s what most people are missing. This isn’t just a Claude story. AMD had 50+ concurrent sessions running on one tool. Their entire AI compiler workflow was built around Claude Code. One silent update broke everything. That’s vendor lock-in. And it will keep happening. → Every AI company will optimize for their margins, not your workflow → Today’s best model is tomorrow’s second choice → If your workflow can’t survive a provider switch, you don’t have a workflow. You have a dependency The fix is simple: stay multi-model. → Use tools like Perplexity that let you swap between Claude, GPT, Gemini in one interface → Learn prompt engineering that works across models, not tricks tied to one → Test alternatives monthly because the rankings shift fast Laurenzo said it herself: “6 months ago, Claude stood alone. Anthropic is far from alone at the capability tier Opus previously occupied.” Never let one vendor own your productivity.

@gyarados__AI がんばれ〜

Claude Codeの基礎を学んでいるんだけど、聞いた事もないカタカナが多くてハゲそう、、、今日中にインストールまで辿り着けるんやろか。もう中学生にも分かるようにChatGPTに教えてもらう

@izutorishima いや本当に、ADHD傾向が加速して寝れません

AI 界隈の ADHD 率の高さをひしひしと感じているが（周囲の ADHD マンがウッキウキで Claude Code 並列稼働して何か作っていたりするため）、逆にこれだけ情報量とトレンドの移り変わりが早く使いこなせばなんでも作れちゃうのと ADHD 特性が奇跡的にマッチしすぎてる生存者バイアスがデカそう

@AnthropicAI When will they finally fix the issue where the permission dialogue box keeps popping up in bypass mode?

New on the Engineering Blog: How we designed Claude Code auto mode. Many Claude Code users let Claude work without permission prompts. Auto mode is a safer middle ground: we built and tested classifiers that make approval decisions instead. Read more: anthropic.com/engineering/cl…

@1Umairshaikh In my recent project, it’s been about 20 hours. I’ve turned into a mad architect.

As a founder how many hours are you actually working per day?

@bloggersarvesh CLAUDE のポテンシャルは計り知れない

call me super annoying but..I will keep repeating this… Claude + SEO is going to make more millionaires in 2026 than Wall Street has in the last decade. don’t bookmark this if it crosses your timeline. just paste this entire thing into Claude.  thank me later.

@cgtwts Gemma は何に使うのが最適だ?

Google Gemma is insane. you can run it locally with OpenClaw in just 3 simple steps: > install Ollama > download the Gemma model > launch OpenClaw using Gemma and just like that, you’ve got a private AI agent running entirely on your own device.

Notebook LM初めて使ってみたけどおもしろい！

GitHub上に公開しているのは、エンジンそのものではなく、この強力な検証プロセスが「何を問い、何を発見し、どう設計したか」という純粋な軌跡のみ 。 このアーキテクチャの論理性について、表面的な議論ではなく真価を共に問えるエンジニアや研究者と繋がりたい 。 詳細・フィードバックはGitHub Issuesまで。 Status: Awaiting Input _ @yokotin34 #SpecLab github.com/frandle331-yh/…

目指すのは、仕様（Spec）と実装（Implementation）の間にある「意味的ギャップ」の完全な消滅 。Syntax CheckLogic VerificationSemantic AnalysisConvergence Testingこれら4つのステージを自律的に無限ループし、論理的破綻を徹底的に暴き続ける 。 AIは「アシスタント」から、妥協なき「監査役」へシフトする 。

現在のAIエージェントは「優秀なイエスマン」という致命的な罠に陥っている 。 形式的なテストはパスするが、設計の本質的な意味が欠落したコードを量産し、後に致命的な破綻を招く 。 この迎合性を物理的に遮断し、意味的差異をゼロにする自律型検証エンジン「Spec-Lab」を構築した 。#LLM #AIアーキテクチャ #CLAUDE

@aryanlabde Just built an AI-native OS layer — a multi-agent LLM beast that takes any spec and relentlessly converges it into perfectly semantically identical code. This thing feels alive. Core is closed, but the full madness is public → github.com/frandle331-yh/…

Vibe coders, what are you working on this Sunday? Pitch your product, get some eyeballs.

Neither AST nor behavior-level — it's semantic/requirement-level. The Auditor reads the spec requirements and the implementation artifacts, then judges whether each requirement is meaningfully satisfied. Not "does the code parse" or "does it run correctly" — but "does this implementation actually express what the requirement intended." DIFF = requirements where the answer is no. It's closer to what a senior engineer does in code review than what a linter or test suite does. The tradeoff is it's LLM-dependent, so you need an anti-yes-man gate to prevent the Auditor from rubber-stamping its own outputs.

What are you building this weekend? Drop your project URL Let’s drive some traffic

@mchulet Super interesting question! I'm already running an "AI as the abstract thinking layer" setup locally with Planner/Designer/Auditor agents + Executor that directly turns thoughts into code/design fixes. It's converging spec and implementation with multi-agent loops and it's genuinely fun to play with every day. Called it Spec-Lab — just commit logs for now, but the semantic diff is getting really close to zero: github.com/frandle331-yh/… This feels exactly like building a new kind of AI-native OS / Agent Operating System. The era where Claude (and other LLMs) can handle this much is wild. Anyone else experimenting with this?

Twitter is cool. But it's 10x better when you connect with people who code. If you're into tech, Al, or programming, let's connect 🚀🚀

@TTrimoreau github.com/frandle331-yh/… AI Operating System

Hey founders ! Looking to connect with people building in: 🍽️ SaaS 🚀 Tech 📲 Automation 🧠 AI tools 📱 Product Development 🔥 Web APP 💻 Devs Drop what you're working on during weekend 👇

エージェントが勝手にスコープを狭めすぎること。 「本当はここまでの権限は与えたくない（またはもっと含めて欲しい）のに、エージェントの判断範囲内でしか開発する前提」なのが個人的にネック。 CursorやClaude使ってる開発者さん、どう感じてますか？ 便利派 vs 使いにくい派、教えてください👀

最近PLANモードの話題がXでちらほら見かけるので、率直な感想。 正直、ちょっと使いにくいと思ってます。