Plan agent in VS Code

The plan agent inside of VS Code collaborates with you to create a detailed implementation plan before you execute upon it. This ensures that all of your requirements are met before any code changes. Let's take a look. Let's

code changes. Let's take a look. Let's

take a look at getting started with the planning agent. Open up the chat view

planning agent. Open up the chat view and select plan from the agents dropdown. Then go ahead and enter a

dropdown. Then go ahead and enter a highle task such as a feature refactoring or bug that you want to plan and submit it. I want to add a few new features here to my badge creation website to add more badges like stars,

licenses, docs, and more. And the

planning agent is getting to work. It's

researching the task comprehensively using readonly tools, codebased analysis, and any additional research to identify requirements and constraints such as going out and fetching specific web pages that I've specified to have

additional information. Sub agents are

additional information. Sub agents are used to gather additional context to help build the full plan. And now it's going to go ahead and break down the task into manageable, actionable steps

and then include open-ended questions to any ambiguous requirements.

And when it's finished, a concise draft plan has been presented. I can go ahead and review this. And I can also see that a few open questions were asked of me.

Let's go ahead and answer these. So I

refine the requirements and get an updated plan. Now I can go ahead and

updated plan. Now I can go ahead and continue to refine this plan to get it to exactly my liking. And when I'm ready, I can go ahead and proceed from the plan through the power of handoff. I

can start an implementation or I can go ahead and open the plan in the editor.

I'm going to go ahead and select that option. By deciding to open in the

option. By deciding to open in the editor, it will save the resulting plan into a brand new prompt file that I can then continue to refine later on. So,

let's go ahead and take a look at that prompt file. Now, the full plan is

prompt file. Now, the full plan is available to me. And what's great here is that I can continue to refine it in this markdown format. I can also decide if I want to save this prompt file

directly into my source code so I can then work on it later or I could go ahead and execute and run this prompt in a new or in the current chat. Now at any time if I start other chats I can always

go over into my agent sessions and go back to the planning session. So let's

say I am ready to implement this. I

could go ahead and execute the prompt or I could go down and say let's start the implementation. I am now in control of

implementation. I am now in control of running it directly in the chat or delegating it to an agent such as the GitHub copilot cloud agent CLI or other agents available to me. And if you're

like me, you may be curious how does a planning agent work. If you go into configure agents and tap on plan, you'll see the exact definition for the planning agent. You'll see all the tools

planning agent. You'll see all the tools available to it, handoffs, and the exact step that the planning agent takes to go from idea to deep research to plan and

iteration with you. Well, there you have it. That's the plan agent. Let us know

it. That's the plan agent. Let us know how you've been using it in the comments below. And don't forget to like,

below. And don't forget to like, subscribe, and share. And with that, happy coding.