Hello and welcome!
(Note: This repository is a distilled, cleaner version of the official DeepLearning.AI Spec-Driven Development course files. It has been reorganized to act as a standalone, easy-to-read curriculum!)
If you are trying to build software with AI coding assistants (like autonomous agents), you've probably noticed something: AI is incredibly fast, but it gets confused easily. It's like having a brilliant, lightning-fast assistant who completely lacks common sense and forgets everything you told them yesterday.
How do we fix this? Through Spec-Driven Development (SDD). In this course, you'll learn how to write clear, authoritative specifications that act as a "Constitution" for your AI.
Think of this repository as your syllabus. We will cover seven core lessons:
- Lesson 1: The North Star (Mission and Vision)
- Lesson 2: Setting Boundaries (Tech Stack Specs)
- Lesson 3: Bite-Sized Assignments (Roadmap Planning)
- Lesson 4: How to Talk to Your AI (Prompting Guides)
- Lesson 5: Ground Rules (Agent Guardrails)
- Lesson 6: Common Pitfalls (and How to Avoid Them)
- Lesson 7: Architecture That Can Be Enforced
The goal is not merely to produce more Markdown. A useful specification connects each important decision to an enforceable mechanism and each mechanism to behavioral evidence. It also keeps the project's authoritative documents consistent as the product evolves.
For repository-level instructions, adapt the copy-ready AGENTS.md template to your project. Use the spec-integrity review worksheet before approving substantial or cross-cutting specs.
Let's get started with Lesson 1!