man I want to know where their creativity comes from, it's like they've built an entire world with a story... but it's just a (highly regarded) collection of packages
Bubble tea is a library for handling TUIs using the Elm architecture, often abbreviated TEA: a model type, a message type, an update function that refreshes the model and emits side effects, and a view function that renders the UI given the model state. The article link goes to the releases page, but the top level README of the project has much better info.
Somehow this whole ecosystem of tools always gives me a bad vibe, and I can't quite pinpoint why.
All the demos and videos are applications with lots of stacked pop-ups/modal windows, and things moving around continuously. It all reminds me of what we typically see in computers in TV shows or sci-fi films.
It just looks like a chaotic mess of things, and I get this really strong urge to just stay away from it all.
Yeah, similar feelings here. I grew up in the BBS world in the 90s, and love a quality TUI experience, but something about this toolset just gives me the ick. I can't keep any of the naming straight; to me it all reads like "combine Chapstick with Cotton Swab inside of Matcha Latte" and my eyes instantly glaze over.
How does this company make money? Is it all just a ZIRP fever dream?
I agree with you about the general vibe being off putting. I’ve been using their libraries for a while now and have to say they are pretty solid though. The terminal components work reliably, and have less UX bugs than the alternatives.
Maybe it's just pattern recognition misfiring, maybe I'm too just used to workhorse software with websites that look stuck in 1999, but everything coming out of that company (Charmbracelet, Inc. according to website footer) feels like it could suddenly get monetized and enshittified next Tuesday morning.
There is also something weird about this entire aesthetics. They all got this 2020s startup vibe of smooth gradients, bisexual lighting, infantile mascots with cartoon or anime inspired styles, vaguely Asian and vaguely feminine. It feels predatory not in the Big Cat with Bloody Mouth way, but in the Cocomelon sensory videos for ages 1 to 3 way. Am I crazy or does anyone else feel similarly?
I would be fine with a chaotic bubbly mess of an outside presentation, if the libraries were more robust and foundational. At the moment the underlying code, when you scratch the surface, have the feel of things thrown together to be replaced at later date.
I bounced off of bubble tea not because of the aesthetics and the unhelpful naming, but because of the programming model: a MVC-architecture cribbed from the Elm language. Why? It completely takes over and rips apart my CLI structure. A CLI is not a DOM or System.Windows.Forms, MVC is scattering around logic and adding indirection layers needlessly.
I am still using huh? and vhs, but their libraries have the feel of looking really good in demo and in the provided examples, but break down quickly when coloring just outside those intended lines.
I've been building a coding agent (https://github.com/abrinsmead/cogent) on the previous version of bubble tea for the past few weeks and it has been nice to work with (though honestly I'm not touching much code).
The biggest blocker I have is that I haven't been able to simultaneously support both mouse wheel scrolling and the ability to select text for copy and paste. I understand that this is a limitation of pretty much all terminals, but we have seen it solved in Claude Code. Maybe this new version has a solution.
Surprised huh v2 isn't included in this and there's no mention of it in the release announcement at all. Quite a few of my apps mix bubbletea/bubbles and huh in different parts of the app, and while they're typically separate and technically could coexist no problem, it would feel a bit weird. Plus huh is advertised as "can be integrated into a Bubble Tea application"; I'm not using it personally but it's surely used by some users. Anyone from charmbracelet here could comment on the situation?
My favourite library from these folks is gum (https://github.com/charmbracelet/gum). The basic premise is simple - instead of using hardcoded variables or in addition/instead of using CLI flags, call gum and capture the STDOUT to get the selected input value(s). Great for turning a bash script into a TUI, uses these libraries under the hood.
I find the pattern of showing interactive TUI if required options/flags are omitted much nicer than showing an error/help output.
I love this design language to death. I know a lot of engineers prefer a no-frills, straight to the point readme (as reflected in these comments), and I get that. But I also don't want to live in a world made out of nothing but boxes.
It feels a bit like visiting Fallingwater and complaining that there are no arrows pointing to the bathroom.
Stoked to see this! I’ve been using bubble tea and all its accoutrements on various little hobby projects for the past few years. Love the ergonomics and aesthetics and can’t wait to try out v2!
Theory: terminal apps are closing the agent self-improvement loop because agents can use TUIs more easily than web/desktop/mobile.
Anomaly, which builds OpenCode + OpenTUI), is also doing some really interesting stuff in this space with their custom renderer. And then there's Ink (https://github.com/vadimdemedes/ink) which is what Claude Code uses. I also built Ink Web (https://github.com/cjroth/ink-web) to make Ink work in the browser.
You know what's even easier for AI agents to use than TUIs? CLIs.
My experience has been that agents suck at using TUIs, and are good at using CLIs. I would argue that agents are a reason that TUIs might die in favor of CLIs.
> the terminal is the most powerful way to interface with the operating system.
There's something about the UX here, how we're interfacing in an abstract way directly with the computer, now with agents. I hope we can get some interesting design study insights over the coming months and years.
I don't see a UI replacing the experience we get with TUIs. Also, UIs seem very clunky, in relation to TUIs performance.
This website is a bit too hyperactive for my taste, and I can't imagine how they hope to make enough money from it to satisfy the VCs. But Bubbletea really is a great TUI and it's properly open source - so I'm happy to enjoy it while it lasts.
I'm with you on the Turbo Vision love, though. Honestly, it felt we were getting to such a good place in the early/mid 90s with beautiful, well-thought-out TUIs becoming the norm.
In retrospect, the universality of basic curses/ncurses on Unix/Linux (together with the death of DOS) was a real a step back - trapping us in a local pessimum for far too long. TUIs languished, and most of us moved to a mix of CLIs and GUIs.
(There's an MIT-licensed port of Turbo Vision available now at https://github.com/magiblot/tvision - pity that it's a couple of decades too late!)
49 comments
[ 2.6 ms ] story [ 69.0 ms ] threadFor a related library in Rust check out Iced: https://book.iced.rs
All the demos and videos are applications with lots of stacked pop-ups/modal windows, and things moving around continuously. It all reminds me of what we typically see in computers in TV shows or sci-fi films.
It just looks like a chaotic mess of things, and I get this really strong urge to just stay away from it all.
The general vibe I get is "script kiddies trying too hard".
How does this company make money? Is it all just a ZIRP fever dream?
Maybe it's just pattern recognition misfiring, maybe I'm too just used to workhorse software with websites that look stuck in 1999, but everything coming out of that company (Charmbracelet, Inc. according to website footer) feels like it could suddenly get monetized and enshittified next Tuesday morning.
There is also something weird about this entire aesthetics. They all got this 2020s startup vibe of smooth gradients, bisexual lighting, infantile mascots with cartoon or anime inspired styles, vaguely Asian and vaguely feminine. It feels predatory not in the Big Cat with Bloody Mouth way, but in the Cocomelon sensory videos for ages 1 to 3 way. Am I crazy or does anyone else feel similarly?
I bounced off of bubble tea not because of the aesthetics and the unhelpful naming, but because of the programming model: a MVC-architecture cribbed from the Elm language. Why? It completely takes over and rips apart my CLI structure. A CLI is not a DOM or System.Windows.Forms, MVC is scattering around logic and adding indirection layers needlessly.
I am still using huh? and vhs, but their libraries have the feel of looking really good in demo and in the provided examples, but break down quickly when coloring just outside those intended lines.
The biggest blocker I have is that I haven't been able to simultaneously support both mouse wheel scrolling and the ability to select text for copy and paste. I understand that this is a limitation of pretty much all terminals, but we have seen it solved in Claude Code. Maybe this new version has a solution.
Why is it so hard to figure out what any of this does? I just want some screenshots
(That said, I do appreciate the artistic flair that went into their website.)
I find the pattern of showing interactive TUI if required options/flags are omitted much nicer than showing an error/help output.
It feels a bit like visiting Fallingwater and complaining that there are no arrows pointing to the bathroom.
https://github.com/weedonandscott/trolley
Anomaly, which builds OpenCode + OpenTUI), is also doing some really interesting stuff in this space with their custom renderer. And then there's Ink (https://github.com/vadimdemedes/ink) which is what Claude Code uses. I also built Ink Web (https://github.com/cjroth/ink-web) to make Ink work in the browser.
The virality of OpenClaw and Claude Code has me wondering if terminals could actually go mainstream (eg used by non-tech users). More thoughts here: https://www.cjroth.com/blog/2026-03-05-terminals-are-cool-ag...
My experience has been that agents suck at using TUIs, and are good at using CLIs. I would argue that agents are a reason that TUIs might die in favor of CLIs.
There's something about the UX here, how we're interfacing in an abstract way directly with the computer, now with agents. I hope we can get some interesting design study insights over the coming months and years.
I don't see a UI replacing the experience we get with TUIs. Also, UIs seem very clunky, in relation to TUIs performance.
This website is a bit too hyperactive for my taste, and I can't imagine how they hope to make enough money from it to satisfy the VCs. But Bubbletea really is a great TUI and it's properly open source - so I'm happy to enjoy it while it lasts.
I'm with you on the Turbo Vision love, though. Honestly, it felt we were getting to such a good place in the early/mid 90s with beautiful, well-thought-out TUIs becoming the norm.
In retrospect, the universality of basic curses/ncurses on Unix/Linux (together with the death of DOS) was a real a step back - trapping us in a local pessimum for far too long. TUIs languished, and most of us moved to a mix of CLIs and GUIs.
(There's an MIT-licensed port of Turbo Vision available now at https://github.com/magiblot/tvision - pity that it's a couple of decades too late!)