Weekly deep dives on AI agents, coding tools, and building with LLMs - delivered to your inbox.
Free forever. No spam.
Subscribe Free
New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.
Introducing Augment's Task List Feature: Take Control of AI-Coded Projects In this video, discover Augment's new task list feature designed to solve the prevalent issue of excessive and uncontrolled code generation by AI coding assistants. I demonstrates how Augment, integrated within VS Code, offers a step-by-step task plan for implementing features such as authentication, allowing users to review, edit, and execute each task. See how Augment's task list breaks down complex problems into manageable steps, enables editing of individual tasks, and can integrate with tools like Jira or Linear. This powerful feature helps maintain control over project changes and enhances coding efficiency. Watch to learn how to leverage Augment for more structured and manageable coding processes. #AI #Coding #programming Links: https://www.augmentcode.com/ 00:00 Introduction to Augment's New Task List Feature 00:40 Overview of Augment and Its Capabilities 01:04 Demonstration: Adding Authentication 02:12 Task List Flexibility and Customization 03:40 Executing and Testing the Tasks 06:13 Additional Features and Integrations 07:52 Conclusion and Final Thoughts
--- type: transcript date: 2025-08-05 youtube_id: ML_29QtcgXc --- # Transcript: Introducing Augment's Task List Feature: In this video, I'm going to show you Augment's new task list feature, which solves one of the biggest problems in AI coding assistants. They generate way too much code without actually giving you control over the process. You've probably been in the scenario before where you ask an AI to add something like, say, authentication to your app or add a new feature and within seconds, it's going off and it's creating dozens of files, implementing features potentially that you didn't even ask for, and sometimes even restructuring the core logic of your project. Now, the task list feature is really aimed at solving that type of issue. Instead of immediate code generation, what it will do is it will create a step-by-step plan that you can review, edit, and execute one by one. In this video, I'm going to show you augment within VS Code, you can access it from the extension marketplace. Alternatively, you can go and just download it from their website. Now, just a brief overview of Augment itself. Within here, we have access to agent chat as well as remote agents. We can send off parallel requests to the remote agent if we'd like where it will run within the cloud for different tasks that we want to ask of it. In this video, I'm going to be focused on primarily the agent capability. What I'm going to ask for is something relatively broad and that can be interpreted in a ton of different ways. I'm going to say let's add in authentication to my application. I'm going to kick off that task. Now, the first thing that Augment is going to do is it's going to perform the retrieval of what's within our current project. It's going to recognize that there isn't any form of authentication set up within the project. Because of that, what it's going to do is it's going to create a task list of all of the component pieces that are involved with setting up authentication within our project. Just to go through what the agent has done, I see you have a fresh next.js application with TypeScript, Tailwind CSS, but there's no existing authentication setup. I'll help you add a comprehensive authentication system. Let me create a plan for implementing authentication using modern best practices. Within here, what we'll see is it's created a task list for itself. And within the task list, we have set up the authentication infrastructure, create authentication components, implement authentication middleware, create protected pages and API routes, add session management, and finally test authentication. So here is the key piece of this. It says now let's start implementing the authentication. I'll begin with setting up the authentication infrastructure with next off. Now let's say for instance, I don't want to leverage next off. Maybe I want to use something like Clerk or maybe I want to use the authentication mechanism within Convex or Subbabase or whatever it is. If I wasn't actually to specify or have the moment to reflect, here in lies the issue that the task list feature aims to solve. Instead of just having an agent go off and do a hundred things and then circle back to you with something that hopefully works, what this is going to allow you to do is to give you a little bit of foresight in terms of the next subsequent steps of the pieces of whatever you're asking for. It allows you to break down complex problems into manageable sequential steps. Within the thread, we have the ability to switch to the task view where we can see it work through all of those different tasks. And additionally what you can do within here is if I expand all the tasks out let's say instead of leveraging next off I wanted to use superbase directly for instance and use the superbase SDK let's say what I can do at any point is I can go and edit any of the task now additionally what you can do within here is you're going to be able to see all of the different subtasks say for instance I don't want it to test the authentication flow because maybe I plan on doing that myself I can just remove that task and it's very flexible in how you want to leverage it. Now, you do have the ability to enable auto modes if you do have a lot of confidence in what you're asking of the agent. You can turn that on and you won't need to approve everything. But in this case, I'm going to approve everything just to make sure that it is continuing on the path and it doesn't go in a direction that I don't want it to go in. Now, the agent is going through and it's creating a number of different files here. It's creating all of the relevant React components for our application. We have a signup form, a logout form, a login form. We can see it's set up our middleware. We can see it's going through and it's setting up all of the relevant pages for authentication. We have the login page, the signup page. We also have an example of a protected page within here. It's even putting in a dashboard page where it will just redirect users if they're not actually logged in. We are going to have a demonstration of a protected page within here. And then finally within here, it's asking us to run our application. Now, the one thing that we will have to get is all of the relevant environment variables from Subabase. But the nice thing with the agent is you can run the process within here. And if there are any errors when it does actually run this, we should see it doesn't have the relevant information like it needs from Superbase, it will actually have the context where it will circle back to us and let us know that we do have to set up that final piece of getting those environment variables. And then finally, what the agent is going to do is it's going to check the task list, make sure everything is complete, and then once it's determined that it is complete, it's going to go through and give us a summary of everything that is done. I'm not going to read through all that we just saw it do, but the one key piece of this that I do like with the platform is it will give us next steps. In terms of the next steps, it says go and create a Superbase project. Update the&v.local with your actual Superbase credentials. Configure the authentication settings within your Superbase dashboard. And then finally, go and run npmdev to test the complete authentication flow. So now I've spun up a quick superbase instance just for demonstration sake. I'm going to go to app frameworks. I'm going to copy the environment variables and then I'm going to put them within our ENV. I'm going to start our development server. I don't see any errors here. And so here is the example application that it's set up for us. And now if I go and sign up here, for instance, I see it's creating an account and we see check your email for the confirmation link. And now if I check my email, I see I have this confirmation for our signup. I'm going to confirm the signup here. And here we go. It did work. Everything is working. Now there is a double navigation which is a simple fix. We can go and we can update that. But within here, the key piece of this is the example of the protected route that it created. And now, just to quickly demonstrate the protected route, if I open up an incognito browser, I can go to the dashboard page. I can see that it is correctly sending us to the login page. So, I won't be able to access anything within this page here. And then in terms of removing the double header here, if I go over to our dashboard page here, and if I just remove this navigation element, and I save that out, then just like that, we have our navigation issue resolved. And then from here you can go and you can iterate through your application. Now just to show you a few other aspects on how you can leverage task list. What you can do within augment say for instance I want to add in a hero area into the dashboard section. Now what I can do within this is I can send this directly into the agent. And if it is determined to be a complex task like you saw when I asked for authentication it's going to go and it's going to break up that task. But since this task is going to be relatively simple we have a couple options. I can send it into the agent directly. Or alternatively, what I can do is I can add this task to our task list. Now, if I go over to our task list, I can see all of the tasks that are complete as well as that new task that I'm queuing up here. And where this can be helpful is what I can do is I can specify to say something like, I want to add in a pricing page. I'll add that to our task list. I want to add in a profile page that is a protected route. I'll add that to our task list. What you can do within here is you can spin off each task one by one if you'd like. if you want to have a little bit more granular control over what the agent does. And this is something that I do see as a common complaint is as projects get more and more complicated, you do have to be a lot more mindful about the changes that are occurring because if you just spin off the agent and ask it to do something, a lot of different platforms, if not really optimized, can definitely go and create a bit of a mess in terms of your projects. And the really nice thing with this is you're going to be able to structure everything that you have in mind. But additionally what you can do with this is you can also connect it to something like Jira or linear where you can take a ticket from linear and the agent can determine okay is this particular ticket something that I need to break up into individual tasks. What it would do is just like you saw here it will go and it will split those tasks apart just like you saw within the authentication feature. But overall that's pretty much it for this video. I really encourage you to check out augment. It is probably one of the most elegant solutions out there. If you found this video useful please comment, share and subscribe. Otherwise, until the next
Technical content at the intersection of AI and development. Building with AI agents, Claude Code, and modern dev tools - then showing you exactly how it works.
