Автоматизація рев'ю коду з Cursor AI: як заощадити на підписках та прискорити розробку

Let's learn as simple as possible what on earth a cursor automation is and how we can start leveraging it in a workflow. Therefore, by the end of this video, you're going to find easy plug-and-play cursor automations you can do for your software or your app that you can start using right away. Does that sound good? Let's jump in. Welcome back, y'all. First off, if you want to access this, just go to cursor.com. Make sure to log in and then you'll notice in your sidebar here automations. This is pretty cool stuff here from Cursor's team here. So, for a couple things we got that's are standing out to me is the ability to find vulnerabilities. As I know a lot are always worried about security. And what I'm going to do in today's video is I'm going to show you step by step how we even set up one of these automations, but on top of that, see it work live. Therefore, a really big puzzle piece for all of this is making sure that you use something like GitHub. If you don't even know what a GitHub is or you want to learn about GitHub, let me just type in Corbin Brown GitHub, watch my videos, or just learn what GitHub is. This is version control and is important for any real software development. Because as you'll notice, a lot of these automations are associated with some type of PR system. But also, you got other cool stuff like fix bug reported in Slack, summarize changes daily, everything of this nature. So, you have two options. Numero eins, we can create a custom one, which we'll do together, or alternatively, use one of the pre-made ones by Cursor's team. So, for example, find critical bugs. We can click this, and what it'll do is that every day at 4:00 p.m. PST, that nice little cron job, it will look at your relevant codebase that you've integrated within Cursor and find critical bugs. Make sense, right? What we're going to create today is going to be a automation that every time I do a pull request, it looks for high-risk bugs. Create automation. High-risk bug. You notice with any type of automations that we have the ability to set different types of triggers, whether it's a cron job, we schedule it every day, every month, every week, or alternatively, you'll see native actions for GitHub, which is probably going to be what you do. So, what I'm going to do is going to be pull request opened. And then we simply give instructions. First off, you're going to choose your relevant model that you want to follow the instructions. Obviously, if we're going to do very high-level things and important things that you may have missed in your workflow, use a high-level model. I do like GPT 5.4. This has been giving me some killer outputs, especially when it comes to like diving deep in the code. I'm going to go with these simple instructions, but they can get pretty deep. Just simply click through Cursor's pre-made ones and you can see how nuanced it can get. Review this PR and audit the code to ensure there is no high-level security risk. Another thing you can attach to these automations is tools, so memories via the chat, or alternatively, MCP servers. The MCP, maybe you are creating a new PR for a Stripe integration, you would integrate Stripe's MCP there. With all that done, we want to select our relevant repo as well. So, I'm going to do one off, and then here, which is really cool, is you can actually set these automations for specific people in your team. So, maybe there's a specific person that just keeps giving you a Stop the AI slop. You can target that person or specific people within your team. For me, I'm just going to say anyone and hit create. Fundamentally, how do we think of these and how you should think of them? Any type of manual labor that you would have typically done in a PR review, like maybe you know from your past experience when it comes to a specific type of architecture within AI coding, it keeps failing, then you would identify that within the automation and then say like, "Really check out whether or not XYZ happens in the code, or is there bloat and dead code being added relative to the main branch?" Like a dead code, bloat code type of automation would be super cool as no one likes it when you could have done a task in 100 lines and it gave you 250 lines. With this on, we can go to our automations here. You'll see total automations is one. Thanks. You'll see run history, and you'll see it here. So, let's actually cook. This is my local app, the one-off app where I just create a bunch of one-off little apps that's open-source code for me to do cool things. And as we know with aigenic coding, we always want to start in plan mode, high-level models. I'm going to say, "Want to add a mini map here for me to create." Yes, I know the GIFs or JIFs, whatever you want to call it. Start a new branch for this, and when it's ready for review, open a PR. I'm going to hit enter here. Let's create a plan. Let's let it go step by step here. Understand the app's architecture and codebase. And then I'll hit execute, and this will from dot zero create a new branch, code in that new branch, open a PR, and then on top of that, we have the automation layer that's going to check the PR for relevant high-risk security bugs. But as you know, and as you can probably assume, there is a bunch of other stuff you can automate within your relevant developer workflow. Like another good example of this I can think of is when I was running teams and interns, a lot of times when they would open up a PR, they would have to DM me directly on Slack to let me know the PR existed. Something like this, they can set up an automation where it's like, "Hey, I opened a PR, automatically DM Corbin." And as we know, we can always view said plan, see if it looks good. FFmpeg, we all know FFmpeg. And I will simply hit build. And from here on out, the entire process I just described is about to happen live in front of us. So, let me go ahead and let this generate. We're going to go ahead and then walk through what this even looks like in GitHub and Cursor automation. And as you can see, right off the bat, it is created and switched to the new branch, feature GIF creator. So, it has finished our relevant plan here. And what I could have done is maybe be a little bit more specific, so it automatically opened this PR. But if we come back over to GitHub, you notice that it's right here, and I hit compare and pull request. Once I click this, and obviously, create pull request, the automation from Cursor should click in. And there we go. Notice two things. First thing, it's automatic, as the name says, but also, it is the very specific automation of high-risk bug that we set together. This is super cool and really cool for two major reasons. Reason number one, Cursor already has a bug bot feature, so you check that out, and bug bot kind of does a similar thing that we're doing here, but you can in theory create your own version of bug bot. And number two, there is a ton of different softwares where literally their entire use case is to be the bug bot. Like, "Hey, we're cleaning the AI slop." In theory, if you're experienced enough developer, you can create your own version of that, and then plug and play. You'll also notice that within the dashboard here, I can come to run history, and you will see that it's currently running. It gives you status, everything of this nature. But that officially goes over Cursor automations. So, as you already know, these style videos, make sure you like, it's completely free. I'll see you in the next. You know what? Corbin quite literally just saved me a ton in software subscriptions, because now we can just literally set up a Cursor automation to do everything we ever wanted in a PR type of video. Super nice.