Mike Cao

@leerob will this be available in AI SDK?

The Cursor SDK can be used to both: 1. Build local, hackable agents with any model 2. Ship products on top of managed cloud agents It handles the harness, sandboxing, computer use, demo videos, GitHub integration, etc. Super cool.

@theo your latest video mentioned a chart you were missing from posthog. curious what is was. maybe we can add it.

Built a simple app so I can keep track of my limits for Claude and Codex. github.com/mikecao/codeme…

@rauchg Yes. I'm literally building every single idea I've ever had, no matter how stupid or impractical, because I want to see it exist.

Some people have been contemplating an idea for years, maybe decades. Obsessing, attempting, discarding, agonizing, retrying. Some of these ideas are unpopular, niche, impractical. Not obviously capitalizable. They live on in the inventor's mind. In 2026, millions of these ideas will come to life thanks to superintelligent coding agents. AI doesn't get tired. It amplifies the individual, and for better and sometimes for worse, it always takes you seriously. "Great idea. Splendid. Wow. You're absolutely right." A world of digital wonders awaits us. This world will disproportionally favor the boldest ideas. Software that once seemed impossible will be one hyperlink away. I can't wait to see it.

oh no, claude is down. this is like having no internet.

Anyone able to use their Claude Pro/Max subscription with Vercel AI SDK?

The @resend template editor UX is terrible. - no reusable components - when you select something to modify, the selection disappears. you have no idea what you are editing - clicking the drag handle opens a menu for some reason and interferes with dragging

My little vibe coded app a month later

@theo no windows? it should be easy enough to do with the electron-builder package.

T3 Code is now available for everyone to use. Fully open source. Built on top of the Codex CLI, so you can bring your existing Codex subscription.

CODEX IS AMAZING github.com/mikecao/audio-…

@thdxr This is the general rule I follow: I give AI 3 chances to solve something. If it can't, I go read and implement it myself.

early in my career when i was learning a new tech or language i would tinker and google whenever i hit a roadblock eventually i realized books had all the information i needed pre-googled for me i think this is happening again with LLMs - sometimes i waste so much time letting the LLM keep taking swings instead of reading something hope the industry doesn't abandon producing good reading material

I have used @WebStormIDE as my primary IDE for years and love it. But with workflows moving to be more AI centric, the lack of native (non-tui) agent support is a deal breaker. And AI chat doesn't count because it requires a separate subscription.

We hit 35K GitHub stars! Thank you everyone for all your support. 🙏 Also, Umami v3.1 is going to be a massive release with tons of new features.

@theo Coincidentally I’ve only run into this context issue with Codex in Opencode and not with Claude Code. 🤷‍♂️

Can't even do basic frontend demos with Claude Code without running into all sorts of weird issues 🙃

I asked ChatGPT, what is the default thinking level for Codex 5.3? And it took 2 minutes to answer (it's medium btw). Bruh, this is your own product, you should know.

I AM HAVING TOO MUCH FUN WITH CODEX

I'm going to let Codex loose on github.com/astrofox-io/as…. It's a massive web-gl and web-audio codebase that needs to be upgraded to a modern stack.

@brobreaux2 I'll have to get Claude to always answer yes to Codex... #agent-ception

@caozilla Same 😭. I just want to leave my agent alone so I can go outside and live my life

Trying out Codex, which model is the best?

First release of my notes app, Noto. Created with Claude Code. github.com/mikecao/noto/r…