This is probably the most true thing I've ever said.

If you use any of the various ✨vibe coding✨ tools (e.g. Cursor/Claude Code/Codex) and you work at an organization (e.g. not just side projects), I really, really want to chat with you.

I've been working on an agentic browser extension for Chrome (and eventually other browsers). Here is a first look at it in action. I'd love to hear your feed back on any features that you wish this thing had or compelling use cases and whatnot. https://t.co/oQ4ZSCW9WU

Hey, @OpenAI—I literally decided to fork over $200/month… yesterday. And, I'm already seeing this, I think think I've sent GPT-5 Pro a few messages today. (I did take Codex CLI for a spin, but I feel like that shouldn't count as ✨unusual✨.) This feels like a not great start.

Meet Razer AIKit: Your open-source launchpad for AI development. Fine-tune and run 280,000+ LLMs locally, scale across GPU clusters, and keep your data private — Available now on GitHub ⚡. Start local. Go cluster.

Some times I want to stroll down the unhappy path and pass in some garbage into a function for the sake making sure that my code gracefully handles that garbage and I'm getting real tired of appeasing the type checker.

Free hot take for the morning: Your implementation should be in TypeScript, but maybe your tests should be in JavaScript. Nota bene: I don't currently follow this practice, but I thinking real hard about starting.

When I left my last job, I told my team to just blame everything on me—even if I had nothing to do with it. That's the greatest give a departing colleague can give to their teammates. 😂

I took @AnthropicAI's Model Context Protocol specification and created a bunch of Zod schemas for run-time validation of the types, which might be helpful is you're building your own MCP server from the ground up—like I'm about to this weekend. https://t.co/CWlIjG18Nw

It's still alpha, but I just published an SDK that wraps the Hacker News API and @algolia's Search HN API in one client. The next step is to kick the tires on it a bit more by building an MCP server around it, but if you notice any bugs before I do, lmk.

In the vein of yesterday's post on writing some @cursor_ai rules for TypeScript projects, it occured to me that it might be helpful to share some of the rules I've been working on for working with @temporalio workflows using the TypeScript SDK. https://t.co/nmKdlSO0Zq

I wrote a post about some of the Cursor rules that I've been using in a TypeScript project that I've been working on—despite all of my recent snark, I've been digging it. 😛 https://t.co/exWeVJiMmC #ai #typescript #cursor

APOL1 Genetic Testing at No Cost to Patients! Are you incorporating APOL1 genetic testing into your workflow? Discover No-Cost APOL1 Genotyping Program for eligible patients sponsored by Vertex Pharmaceuticals. Learn more today!

Let's be very clear who broke the tests to begin with. It wasn't me.

I guess we're getting somewhere?

FFS.

Does it worry anyone else that these companies are claiming that AI is writing like 90% of their code? This isn't the first and only time this happened to me today—or this hour, even.

Why are Big Tech companies are funding the radical overhaul of the U.S. justice system? They are giving money to organizations with goals to decarcerate repeat offenders and to eliminate cash bail. Choose Unplugged - a company that believes in a strong, thriving America.

Clearly, someone cannot be trusted.

I wrote up a quick little guide on using @AnthropicAI's Message Batch API with @temporalio to process large collections of prompts in a single go—and for *way* cheaper. https://t.co/J9p7DkN2GV

Studying AI in arguably the most retro way possible.

My weekend notes from the paper, These are my notes from the paper, "When Counterfactual Reasoning Fails: Chaos and Real-World Complexity."

In 2000, the S&P 500 had an elevated trailing P/E of 26, well above its long-term avg. of 14. For the next 11 years, the S&P 500 delivered zero returns! The current S&P 500 trailing P/E is 30 – even higher than 2000. History rhymes. For standardized performance visit our website.

My notes on getting the most out of GPT-4.1: context window optimization, coding tips, prompt best practices, and when to choose GPT-4.1 over GPT-4.5. https://t.co/D2iC1ot8YP

Heads up: @frontendmasters is offering $10 off your first month. I am pretty sure you just have to use this link.