From the course: Spec-Driven Development with GitHub Spec Kit
Introducing Spec Kit: Spec-driven development with AI agents - GitHub Tutorial
From the course: Spec-Driven Development with GitHub Spec Kit
Introducing Spec Kit: Spec-driven development with AI agents
- Spec Kit was announced in September, 2025 and is currently in development. So what I'm demonstrating here is very much beta software and has some significant issues you'll see as we walk through it. But the overall concept is there and in the blog post that was published when Spec Kit was announced, there's a fairly good breakdown of how Spec Kit works and how to work with it. Now let's take a brief look at what the idea is here first before we dive into the code itself. So when you work with Spec Kit, you're walking through four distinct processes. The first process called Specify or Specify, there is some discussion within the Spec Kit developers themselves and within the community, whether this is Specify or Specify since it's called Spec Kit. And because English is such an ambiguous and ambivalent language, it is not clear how to pronounce this. So we're in a bit of a GIF GIF scenario here. We'll see what the developers end up landing on. But the last I heard in a live stream from one of the lead developers was Spec Kit. So I'm sticking with that for this course. Using Specify you provide a high level description of what you're building and why, so that the coding agent can generate a more detailed specification. The idea here is to leverage the power of the coding agent to build out a specification beyond your specific needs so you don't have to do all the work. The next step, once you have a spec is to run the plan command. The plan command grabs the spec and builds out a step-by-step plan for the coding agent to go through. And the idea here is when you get the system to generate a plan, you go in and edit it to make sure it's doing what it's supposed to be doing in the order you want and that everything is in place. And the way this is built is so that the coding agent will actually follow this plan very carefully. And this is done by including things like checklists at the bottom and some scripts that assist the coding agent in doing its work, which we'll look at later. Once you have a plan, you tell Spec Kit to generate a list of tasks. These are the actual tasks the AI coding agent will perform. And again, you can edit them. And then finally, you tell the coding agent to implement all of these changes one at a time. And there are different ways of doing this. So you have specification, you have planning, you have task breakdown, and you have implementation. And they are discreet steps built so that you are expected to provide input at every step. Now, in addition to this, there is a constitution you start out with that describes the overall environment you want to work in and whatever constraints you have. And there is a lot of back and forth between you and the coding agent as it's working.
Contents
-
-
-
Introducing Spec Kit: Spec-driven development with AI agents3m 9s
-
(Locked)
Setting up Spec Kit4m 3s
-
(Locked)
Creating a constitution for your project3m 30s
-
(Locked)
Generating a project spec with the /specify command4m 6s
-
(Locked)
Reviewing and updating the spec with human-in-the-loop3m 13s
-
(Locked)
Generating a project plan with the /plan command3m 14s
-
(Locked)
Understanding and reviewing the project plan7m 8s
-
(Locked)
Generating and implementing tasks using the /tasks command5m 36s
-
(Locked)
Understanding where Spec Kit is most useful4m 51s
-
(Locked)
Exploring how Spec Kit works and how you can modify it4m 13s
-
(Locked)
Contribute and shape the future of AI coding tools2m 13s
-
-