Tommy Nguyen

@steipete My wishlist: - Pinned left sidebar: having to open the sidebar, switch sessions, and close the sidebar again is painful. - Command palette: cmd + k would make a whole lot better experience. - Native Mermaid chart rendering: super useful for plan mode.

@tuannvm Yeah, hard to prioritize when there's so few users.

That's about as many sessions as my Mac Studio can take.

@HAProxy is one of the few vendors that cares deeply about the actual business value customers can get from using their products. Can’t wait to be back to share even more significant improvements, thanks to HAProxy’s continuous evolution!

@steipete Same complaint for whichever apps try to install global hooks; uninstall immediately.

New pet peeve: cli's that install new skills onto my system without asking.

@daniel_nguyenx @OpenAIDevs SSH is more reliable, good that both Codex mobile and macos support

@OpenAIDevs Black magic. But none of this matters if you don't fix the connection issue :(

Codex anywhere and everywhere, all the time. Now your Mac doesn’t have to be unlocked for Codex to use your computer. From your phone, Codex can securely use apps on your Mac, even when the screen is off and locked. #locked-use" target="_blank" rel="nofollow noopener">developers.openai.com/codex/app/comp…

@OpenAI Is codex spark accessible with the $100/month Pro tier?

We’re updating our ChatGPT Pro and Plus subscriptions to better support the growing use of Codex. We’re introducing a new $100/month Pro tier. This new tier offers 5x more Codex usage than Plus and is best for longer, high-effort Codex sessions. In ChatGPT, this new Pro tier still offers access to all Pro features, including the exclusive Pro model and unlimited access to Instant and Thinking models. To celebrate the launch, we’re increasing Codex usage for a limited time through May 31st so that Pro $100 subscribers get up to 10x usage of ChatGPT Plus on Codex to build your most ambitious ideas.

@what_the_func I guess they have a great memory of exactly where they stopped and resumed from there.

For the terminal multiplex haters (ie tmux/zellij), how are you starting a session on your remote device, taking a break and resuming that session on say another machine somewhere else?

I think the main reason people still stick with Tmux is the ability to restore or resume sessions across devices. The great thing about Tmux is that every time I log in to my existing session, everything is the same, and I end up in the same pane in the same window, so there’s no context switching.

There's such a deep misunderstanding out there about tmux and I get so many absurd issue reports demonstrating that. Many don't realize that using them is like running a Windows VM on your Mac, and complaining to Apple that iCloud sync isn't working from Windows in the VM. They are super powerful and have their use and I am happy to support them in any way I can. I'm not anti-multiplexer, but I wish more people understood the architecture a bit more.

@trq212 @Claude This duplicates with the normal @claude review. Would it be better if there is an explicit handle?

You can now configure Code Review to run manually on demand by typing `@claude review` in your Github PR or comments. This gives you more control to use Code Review on select PRs where the review would be useful.

Released today: /loop /loop is a powerful new way to schedule recurring tasks, for up to 3 days at a time eg. “/loop babysit all my PRs. Auto-fix build issues and when comments come in, use a worktree agent to fix them” eg. “/loop every morning use the Slack MCP to give me a summary of top posts I was tagged in” Let us know what you think!

I'm a big fan of making the most out of solid & well-designed AI workflows, and @cerebras has opened a new frontier!

cli is the new mcp

@what_the_func Unless official Go SDK offers something much more compelling, I don’t see a reason why I have to switch out such a stable and reliable implementation.

An official golang SDK for MCP has existed for months now yet people still use and contribute to mcp-go. I'm honored and grateful but just curious why some people haven't migrated?

Happiness is when you turn on "don't ask" mode and can just go on and on with no errors, no disruptions. Everything just works. Thanks to @bcherny 's best practices x.com/bcherny/status…

Happy New Year to all builders. What do you expect or predict for the future of LLMs in 2026?

I also asked 3 frontier models for their predictions.

Human interaction will become increasingly valuable in an AI-fatigued world. As AI-generated content becomes more common, people will likely place greater value on human interaction and service. There's something irreplaceable about talking with someone you trust, someone you can meet face-to-face.

Model inference will become cheaper, while the intelligence of LLM models will improve. This is happening because companies are finally seeing returns on their infrastructure investments and finding ways to make training and serving more efficient. These improvements will lead to better prices for customers, and the money saved can go toward developing even smarter models (AGI?)

Here are some of my quick thoughts and predictions: - The focus of AI applications will shift away from the model itself. Instead, what truly matters is how you organize your tools, structure your workflows, and design your overall system.