Visual Studio Code, featuring Gwyneth Peña-Siguenza, guides developers through prompt-driven development and AI-powered codebase refactoring using GitHub Copilot with a FastAPI + Cosmos DB project.

Refactor an Existing Codebase Using Prompt-Driven Development with GitHub Copilot

Featuring: Gwyneth Peña-Siguenza (@madebygps)
**Produced by Visual Studio Code **

This session demonstrates how to leverage GitHub Copilot to refactor an e-commerce inventory management app written with FastAPI and Azure Cosmos DB. The walkthrough covers recognizing code smells, building contextually-rich prompts, running Copilot in Agent Mode for refactoring tasks, and reviewing AI-generated improvements step by step. The approach emphasizes best practices for integrating AI assistance into real-world development workflows.

Video Chapters

• Intro (00:00)
• Getting Started (00:36)
• Choosing Directory & file: product_crud_batch.py (02:25)
• Writing Custom Instructions (03:50)
• Creating Effective Prompts (05:25)
• Running Prompts with Copilot Agent Mode (09:58)
• Testing Changes (19:11)
• Summary & Key Takeaways (23:39)
• Wrap Up (24:14)

Key Topics Covered

• Refactoring real-world Python code for an e-commerce scenario
• Identifying and addressing code smells in an inventory system
• Constructing targeted instructions and prompts for Copilot
• Using Copilot's Agent Mode for iterative improvements
• Reviewing, testing, and validating AI-suggested code
• Best practices for making AI-driven refactoring reliable and maintainable

Demo Project

• cosmosdb-fastapi-functions GitHub Repo

Best Practices Highlighted

• Develop focused, stepwise prompts for each refactor target
• Validate Copilot's output through tests and code reviews
• Iterate rapidly, integrating manual and AI changes
• Leverage Agent Mode for continual improvement and prompt experimentation

Connect with Visual Studio Code

• Twitter: @code
• Bluesky: vscode.dev
• YouTube: youtube.com/code
• LinkedIn: Visual Studio Code
• GitHub: microsoft/vscode

For developers looking to enhance productivity and code quality using Copilot, this video offers actionable insights and hands-on techniques that can be applied to similar Python and Azure-based projects.