YouTube Deep Summary

Vibe coding. It's the new trend where people code without really looking at the code at all. These days, almost anyone who speaks English can do it. But Vibe coding comes with a dangerous side effect, and that's vibe debugging. When you're building stuff using these new AI tools like cursor or clawed code, your app might seem like it's running without any issues. And I say might because in reality, that's rarely the case. You often run into bugs. And that's usually because the AI doesn't fully understand what it's writing or it forgets what it already wrote due to context window limitations. So now you're stuck with code that you didn't write, don't understand, and don't know how to debug. Now, how do you debug code you didn't write? That's exactly where Sentry comes in. Sentry can help vibe coders and professional devs debug their code efficiently, even when they didn't write it themselves. It's a platform many real developers already use every day, but they've now made it even more powerful by releasing their new feature. So, what exactly is Sentry? For those of you who haven't used it before, Sentry is a platform that automatically tracks and reports errors in your app. And if we combine the traditional programming workflow with these new AI workflows, that's how you're actually going to build reliable software that doesn't break. Now, you might think you could just log these errors in a MD file or maybe run some bug bots that catch those errors, but that isn't the optimal solution. And you might think, what makes Sentry different? It's actually a proper system for catching all those repeated errors in your app. They've got built-in support for different frameworks like Nex.js or Python or whatever you're using. Sentry tracks how often an error has happened, when it happened, where it came from, and keeps a full history of it, all organized in a clean dashboard. Now imagine this. If your AI agent had access to all that context, every stack trace, every repetition, every log, solving the problem becomes way easier. It could easily go ahead and fix that issue with all that extra context. So, what have they released? It's their new MCP server, and this thing is amazing. Sure, it eliminates the debugging headache, but there's something even more impressive that caught me off guard. Normally, if you were working with Sentry in your project, what you do is go to the website, create a project there, then come back and install the SDK for the specific framework you're using. For example, if you're using Nex.js or Python, you'd install their respective SDKs. After installing the SDK, you'd need to initialize Sentry in your actual code so that when your app encounters errors, Sentry can catch and log them in your dashboard. Now, what's the added perk of this Sentry MCP? Let me show you. You can see that the Sentry MCP is already added in cursor. I'll give you details on what the tools are and how to install the actual MCP server because there's a bit of authentication involved. But for now, just focus over here. First, I asked if it had access to the Sentry MCP server. It replied yes. Then it called the tool who am I to check my user details because as I mentioned, we've already authenticated with Sentry for this MCP to work. It returned who I was, what my organization name was. Then it ran the tool called find organization and it identified that my organization was automata. After that I asked if it could access all of my projects. Remember we usually have to manually create a project in Sentry but at this point I hadn't made any project. So it ran the tool called find projects and saw that there weren't any projects set up in my Sentry organization yet. And this is what actually surprised me. It said it could create the new Sentry project for me and automatically set up the error monitoring for my application. Now this this is just on another level. You don't have to go through the UI or do anything manually again. This is how AI agents, AI based applications and these amazing MCP servers are changing everything. So it said it had helped me and set up the Sentry project for me. It checked if I had any teams then proceeded to create the project for my notes application and of course it knew the command to install the Sentry SDK. In my case it installed the SDK for Nex.js. Then as expected there were configuration files. It set those up as well. After that it modified my code because like I mentioned earlier you've got to update the code so that errors can be logged in the Sentry dashboard. And another thing as I already said the manual processes are completely gone after integrating the code. You typically need to get your DSN from Sentry which is kind of like an API key used to send error logs to Sentry in the cloud. Normally you'd get this manually but here's what happened. Since it was handling everything, it fetched my Sentry DSN, automatically added it to my configuration, and even gave it back to me in case I needed it elsewhere. Again, this is what I'm trying to tell you. It handled all of that automatically. Basically, the MCP gives us these tools that automate the entire Sentry setup process. Then, it actually asked if I wanted to test whether error logging was working, and I said yes, go ahead. So, in the app I used, it added three buttons that would trigger errors. These are the three error buttons you see. Actually, let me just open up the app so you can see how it added those buttons. So, this is the app and these are the buttons it added to test if the Sentry connection was working. If we go into my dashboard and check the issues, you can see the different ones that have come up during testing. So, yeah, the issues are showing up correctly and Sentry has been properly initialized. Now, how do you install the MCP server? It's actually quite easy. before you go on and actually install the Sentry MCP server, you need to make sure you've got a Sentry account. Just open it up. And if you don't have an account, like right now I do, then just go ahead and sign up. You don't need to create a project or anything. Now, after you've signed up, you're going to get this command from the docs. And don't worry, I'll leave those docs in the description below. Just copy this JSON configuration. And if you scroll down, they've actually mentioned the verified clients, which means Claude desktop, Claude AI, which just means Claude Code, Windsurf, Cursor, and also VS Code, and GitHub Copilot. So yeah, it's verified for all of these. They've listed some install instructions, but it's pretty much the same for all of them. Just go ahead. Like for example, in my case, my MCP.json is empty right now. So I'm just going to copy this whole thing. And now you're going to see that the MCP server, the Sentry MCP, has been added. After you add it, it's going to run. At first, it'll appear red, but once it runs the command, it'll automatically open up in your terminal like this. And you'll get this message saying the MCP CLI is requesting access. You just need to approve it. Once that's done, if you go back and it still hasn't turned on, just toggle it off and on. And now you'll see that the tools have been enabled. And we've got all of our tools right in front of us. So, I did already show you that it can automatically initialize a project for you. But what else can it do? They've provided some example calls that give you a pretty good idea. Obviously, you can ask it to list all the issues in your project. You can also ask Sentry about a particular file to check if there are issues in it and then go ahead and fix them. There's also Sentry's own AI powered SEIR which can analyze your issue and provide a solution. Now, this is part of the paid features. So, that's up to you. But even without Seir, Sentry still gives you enough data about a specific issue that cursor or claude code, whichever you're using, can automatically fix it. I just wanted to show you some of the tools the Sentry MCP offers. You can see that many of the tools are related to using Sentry as a platform like Who am I, find organizations, find projects, create DSN or create projects, which help you manage the Sentry platform through MCP like I showed you earlier. You don't have to do any of it manually. Then there are other tools like get issue details, find errors or even begin seir issue fix which help you get detailed insights into the issues or actually fix them. In order for you to purposely see it in action, I purposely put an error in my application. One that completely broke the app and wouldn't let me use it. If we go into my dashboard, you can see that the issue shows up here. If I click on it, you can see that the issues listed and we get all the details about it, including the trace. Using all this information, the LLM can actually understand where the error came from, track it, and get accurate insights on how to fix it. So, first I just blatantly tried using SEIR to fix it. That's when it told me I had to enable it and that it's a paid feature. Then I went ahead and told it to get the details of the issue and propose a fix for me. It called the get issue details tool. You can see we've got the issue ID, our organization, and then as a result, it gave us the issue summary, some context about its occurrences, and even as we scroll down, a trace, which is pretty helpful. It also gave some additional context, which didn't really matter much in this case, but essentially, we gave it the issue details and as you can see, it recognized the type of error. It was a type error. Something was not a function. It proposed that the error should be removed and also suggested an additional check to make sure it doesn't happen again. Then it just went ahead and started fixing the issue. And once I accepted the fix, the application started working again. But how do you fully take advantage from this? If you want to have this full setup with Sentry monitoring your errors and then logging them and tracing them, which will essentially help the agent debug your code better, then you need to make sure that you use it properly and use it from the start. So even if you are building stuff with AI, the agent knows that you are using it and make sure that it sets it up in your project correctly and Sentry is properly utilized. That brings us to the end of this video. If you'd like to support the channel and help us keep making tutorials like this, you can do so by using the super thanks button below. As always, thank you for watching and I'll see you in the next one.