Wrote a skill that runs codex /review in a loop until there's no booboos anymore. Caveat: It won't fix system architecture for ya, so you still need BRAIN as master model. github.com/steipete/agent-sc…

Too many AI coding agents, too many session histories everywhere. Claude Code. Codex. OpenCode. Pi. I built Agents Aggregator — a local web app that puts them all in one UI. Browse, search, and watch sessions stream live as they run. Even send keystrokes back into tmux from the browser. 100% local. No cloud. No auth. Vibe coded in an afternoon. github.com/ptgamr/agents-agg…

With agents writing code I’ve become more bullish than ever on Effect and the problems it solves If you’re not using it, I’m curious on why and what you’re using instead to solve the same challenges - interested to see what the ecosystem looks like

Everybody is adding a feature where you can manage your agents from your phone. Don't use it. You'll just get even more addicted, and will burn out even quicker.

Since we open-sourced pi-autoresearch, @Shopify teams have been running it on everything. Results so far: Unit tests: 300x faster React component mounting: 20% faster CI build time: 65% reduction Made pnpm run faster Autoresearch never stops trying things you'd never have time to try. Repo: github.com/davebcn87/pi-auto…

RAG is broken and nobody's talking about it. Stanford researchers exposed the fatal flaw killing every "AI that reads your docs" product in existence. It’s called "Semantic Collapse," and it happens the second your knowledge base hits critical mass. If you've noticed your AI getting "dumber" as you add more data, this is exactly why. Right now, companies are dumping thousands of documents into their AI, thinking it’s getting smarter. When you add a document to RAG, it converts it into a high-dimensional vector. Under 10,000 documents, this works perfectly. Similar concepts cluster together. But past 10,000 documents, the space fills up. The clusters overlap. The distances compress. Everything starts to look "relevant." It is a mathematical law called the Curse of Dimensionality. In a 1000-dimensional space, 99.9% of your data lives on the outer edge. All points become equidistant from each other. That perfect, relevant document you are looking for now has the exact same mathematical similarity as 50 completely irrelevant ones. The Stanford findings are brutal: At 50,000 documents, precision drops by 87%. Semantic search actually becomes worse than old-school keyword search. Adding more context doesn’t fix the AI. It makes the hallucinations worse. Your "nearest neighbor" search isn't finding the best answer anymore. It's finding everyone. We thought RAG solved hallucinations. It didn't. It just hid them behind math.

*Finally* read through @samwhoo's blog on LLM quantization. It's incredible. For many (even in tech) the understanding of how LLMs work stops at the surface level. Sam is helping us all go deeper, digging into the interesting facets of how AI models truly work. Read it!

I think I finally understand what an agent is. It's a prompt (or several), skills, and tools. Did I get this right?

RT @shl: Gumroad’s test suite of 16,000 tests has been flaky for years. This slowed down shipping tremendously. This week, Gianfranco used…

I packaged up the "autoresearch" project into a new self-contained minimal repo if people would like to play over the weekend. It's basically nanochat LLM training core stripped down to a single-GPU, one file version of ~630 lines of code, then: - the human iterates on the prompt (.md) - the AI agent iterates on the training code (.py) The goal is to engineer your agents to make the fastest research progress indefinitely and without any of your own involvement. In the image, every dot is a complete LLM training run that lasts exactly 5 minutes. The agent works in an autonomous loop on a git feature branch and accumulates git commits to the training script as it finds better settings (of lower validation loss by the end) of the neural network architecture, the optimizer, all the hyperparameters, etc. You can imagine comparing the research progress of different prompts, different agents, etc. github.com/karpathy/autorese… Part code, part sci-fi, and a pinch of psychosis :)

I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit. My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently. So, here goes.

Damn it, @cursor_ai GPT-5 Codex think-hard track down a really hard rendering bug. CRAZY!

Damn, Cursor's composer is too good!

We're proud to announce: SportContract is now part of Hudl! 🏒 SportContract is a hockey-first platform trusted around the world. Together, we’re delivering smarter tools for video analysis, player development, and team communication. More here 👉 tinyurl.com/3nrkrrdr

I mentioned the trifecta of children books What Do You Do With An Idea/A Problem/A Chance on the Lex podcast. The author is Kobi Yamada, illustrated by Mae Besom. Wonderful books for kids (and their parents!). amazon.com/What-You-Do-Matte…

If Chrome just killed your ad blocker, switch to Brave. Brave’s ad blocking is built into the browser itself so it isn’t affected by the Manifest v3 changes that Google is rolling out.