Transcription: OpenAgents Episode 197 - Reverse Engineering Cursor

Source: https://x.com/OpenAgentsInc/status/1989020655602618370 Wiki source: https://raw.githubusercontent.com/wiki/OpenAgentsInc/openagents/Video-Series.md Media title: OpenAgents - Episode 197: Reverse Engineering Cursor The year is 1900 and the wor... Upload date: 20251113 Transcription model: gpt-4o-transcribe-diarize Generated at: 2026-06-01T03:22:44Z

Machine-generated transcript. Review speaker labels and wording before using this as quote-grade source material.

[00:00] Christopher David: How can open agents help me? Well,

[00:02] Christopher David: Claude, you can tell them about last episode. We've got ten upgrades.

[00:06] Christopher David: We're going to do one video per upgrade, starting with number one. What's number one? Claude, tell them what number one is.

[00:12] Christopher David: Execute, execute,

[00:13] Christopher David: let's go.

[00:14] Christopher David: Yeah, number one. Ditch the TUI. Replace janky fake terminals with a proper desktop app featuring a sidebar,

[00:22] Christopher David: chat history,

[00:23] Christopher David: and widgets for managing long-running agents.

[00:25] Christopher David: Hey, hey, hey, TUI people.

[00:26] Christopher David: Why people?

[00:26] Christopher David: How long does it take you to switch to a previous thread?

[00:30] Christopher David: Yeah, remember?

[00:31] Christopher David: Is it continue or resume or last or dash dash or what do I do?

[00:35] Christopher David: How about clicking on it?

[00:37] Christopher David: Oh, you want to be cool about command up and down?

[00:39] Christopher David: Oh yeah, can you do that?

[00:41] Christopher David: You cannot do that. You can do that with open agents.

[00:45] Christopher David: Upgrade number one of ten.

[00:46] Christopher David: More coming.

[00:47] Christopher David: See you soon.

[00:48] Christopher David: The year is 1900,

[00:50] Christopher David: and the world's best investors have poured a lot of money into the company building a faster horse.

[01:02] Christopher David: Congratulations to Cursor.

[01:04] Christopher David: I know that cursors read in our tweets.

[01:07] Christopher David: I know that cursors read in our tweets. So I hope they enjoy this one.

[01:11] Christopher David: That's StarCraft. This isn't fucking StarCraft.

[01:14] Christopher David: Okay.

[01:15] Christopher David: It's good. It's good. It's good.

[01:17] Christopher David: So follow-up from our last video. So ditch the T-U-I.

[01:21] Christopher David: You know, you gotta also ditch your old I-D-E.

[01:24] Christopher David: Okay, so I said something rubs me wrong,

[01:26] Christopher David: I have to say. When you take an open source product,

[01:29] Christopher David: you fork it and you closed source it and you build a bunch of stuff on top. First of all, I think the model's all wrong. But something about that just rubs me wrong anyways.

[01:38] Christopher David: Bro,

[01:38] Christopher David: what do you need $2 billion for?

[01:40] Christopher David: Is that for buying out people like me?

[01:42] Christopher David: Because you're going to need a lot more than that.

[01:46] Christopher David: Okay,

[01:47] Christopher David: so we're going to reverse engineer Cursor.

[01:48] Christopher David: But look,

[01:49] Christopher David: look, look, look, hang on, hang on, hang on.

[01:51] Christopher David: Lawyer GPT says I should say this.

[01:54] Christopher David: In this video, we're exploring Cursor purely for educational interoperability purposes.

[01:57] Christopher David: All the code we produce is entirely original under U.S. law provisions for yada yada yada. Okay, great.

[02:01] Christopher David: So you'll take a look on our open source repo.

[02:06] Christopher David: Uh we got some cool stuff in a new folder called docs slash R_E_ cursor and you know cursors just sitting on your computer. You can just have the agents, like, look at it. Uh so we got a little deep research report about cursor's architecture from stuff you could find online. And then had our coding agents just take a look at like what's going on locally. For example, you can like look at all the node modules that they use, be curious to see someone run uh run this through uh M_P_M_ audit and just see if there's any uh problems there.

[02:33] Christopher David: Ooh thumbs up.

[02:34] Christopher David: Oh, what else we got here?

[02:36] Christopher David: Comparing the dependencies of Cursor versus VS Code.

[02:38] Christopher David: Like what do they add on top of VS Code?

[02:41] Christopher David: How does that map to architecture?

[02:45] Christopher David: Themes, nice themes.

[02:47] Christopher David: I like them. Good job.

[02:48] Christopher David: and what start speculating about like how would we combine some of the local stuff that we've been talking about um not depending so much on their cloud stuff but like what would a more local version of that look like and then um i fed it the transcript of our episode 195 and i was like yeah what happens if we were to just kind of like do our own version of all this

[03:17] Christopher David: What does it say down here? Hang on.

[03:21] Christopher David: Um, what does it say down here? Hang on, hang on, hang on. So the budget's the good stuff. We're not going to go through all this, but like,

[03:31] Christopher David: wow, this is a lot. Very good job,

[03:33] Christopher David: GPT-5.

[03:37] Christopher David: Hang on, there's something I want to show you.

[03:40] Christopher David: Wait. Okay,

[03:41] Christopher David: weeks.

[03:41] Christopher David: Yeah, yeah, yeah, yeah, yeah, yeah.

[03:43] Speaker B: Who wrote this roadmap?

[03:44] Speaker B: Claude?

[03:45] Speaker B: Probably Claude?

[03:51] Speaker B: Once you can spec something out and you got like a good model for it like the creation of software,

[03:58] Speaker B: time and money all that's falling to zero and a lot of people, companies, investors just have not grappled with the ramifications of that.

[04:09] Speaker B: Holy shit. Okay,

[04:11] Speaker B: this is what I was trying to show you.

[04:13] Speaker B: Six weeks to MVP.

[04:14] Speaker B: Twenty weeks to 10x better than Cursor. Guys.

[04:20] Speaker B: Six days...

[04:22] Speaker B: max to MVP.

[04:24] Speaker B: Twenty days to 10x better than Cursor. I don't think we need all that time.

[04:29] Speaker B: See you soon.