Загрузка видео...
Не удалось загрузить видео
Introducing Sundial!! A brand new text editor built from the ground up for working with agents. Here, we're using Sundial to co-write a new Sundial feature.
49,883 просмотров • 3 месяцев назад •via X (Twitter)
Комментарии: 64

Add Claude Code or Codex in a few clicks so they participate in real-time like a human would. Or press cmd+J to chat with Sunny, Sundial's agent.

What I like most about Sundial is the review panel. We're used to text editors with a basic history tab, but given the amount of work we're getting done with agents, we need better tracking and attribution across multiple threads.

Comments are also immensely helpful since they are easier to use than chats for targeted changes.

In Sundial each change is linked to the chat turn that produced it, so agents can build on top of each other's work while you keep the final say on what stays and what goes.

Join us for early access:

niiice! congrats with the release!

💛

@Tocelot Downloading this🏃♂️💨

@Tocelot Yas

So cool, been wanting something like this for a while! it’s great that I can consolidate my token usage to a provider that I’m already paying for vs having to pay for separate services (eg google or notion business account)

congrats on the launch! WE LOVE SUNNY

let's gooo team sundial!

thx jon!

congrats on the launch!

Ty ^^

excited to try!!

🫶🏼

a text editor built around agents from scratch instead of bolted on is the right move, curious how it handles multi file context

We made multifile context native because that is how agents naturally work

that's the part that usually breaks first in other editors so good to hear it was built in from day one

pretty cool, signed up for beta!

Nice

Oh ok! I need to try this

pls do ~

I like agent-native editors, but I’d test the handoff more than the chat. If Claude Code or Codex touches a feature in real time, show changed files, stuck state, and the one command that proves it still works.

would love for u to try :o

will do. the thing I’d poke at is whether the editor keeps the agent’s work inspectable: changed files, failing command, next proposed patch. that’s usually where agent-native editors either become real or turn into a nicer chat pane.

fs

Does this have Linux support?

It's a web app so can support local agents from any machine

Awesome signed up!

yay

ah sorry can't find you on list, dm me and will send beta link!

Soooooo useful

Congrats on launching @belindmo!

Thx ☺️

Very cool

I’m on the wait list!

👀

I’m curious why write with Claude instead of plugining the ai into the application

Ya we have both depending on ur preference ^^

@Scobleizer Very interesting stuff, the amount of code generated by AI is difficult to review and only possible after a huge chunk is already done, especially with goals and loops. Sundial’s presentation seems like it helps with that, would love to try it out.

@Scobleizer fill this, tracking people to onboard here! Would love to add you

👀 @outofbabylon

The biggest problem I'm facing right now is when using cloud code or codex to develop projects - for example, when generating a lot of related documents, session logs, or memo management - I don't have a really good text editor on my Mac.

This is cool! Congrats on the Intro/launch

Ty ^^

@belindmo this looks so neat - I'd love to get access! It will make my handoffs with agents, collaborations with agents, and orchestration with agents so much more seamless! yes, please!

rolling out soon :)

the honest answer is that most text editors aren't designed for working with code that's generating code. sundial's real innovation is in the feedback loop, letting you iterate on the meta-programming itself.

definitely, RSI is the way

and being able to track exactly what happened so while the agents self-improve you know what they did

love the diffing ui :) so thoughtful

That’s from @turboblitzzz c:

Cool 👍

shared undo stack is the part VS Code gets wrong for agent work agent touches 40 places, one goes bad, and you're either reverting your own edits too or manually picking it apart line by line

agreed, it is really bad. the root issue is lack of attribution which sadly is unrealistic for vscode/cursor to support

How did you create the video? What tools did you use? Both for navigation and for shooting and editing (mouse and zoom). Thanks 🙏🏼

How much is the subscription fee

In beta, but probs free forever if u use your local agents 🙂

Great stuff! dropped a dm

great work but it's a fucking markdown editor, I should not have to write an application essay and cover letter to get access to the beta.

includes model credits

add BYOK like a normal human being

