In this two-day hands-on workshop, you’ll take a real-world .NET API built with Dapper and incrementally migrate it to Entity Framework—entirely using GitHub Copilot inside VS Code. We’ll explore all three Copilot modes—Ask, Edit, and Agent—along with prompting strategies, context engineering, and reproducibility patterns.

You’ll learn how to:

Create custom markdown instructions and project settings to control Copilot output
Use prompt-driven Git workflows to incrementally transform production code
Run side-by-side “audition branches” to compare Copilot outcomes
Tune Agent Mode using SQL schemas, code templates, and naming conventions
Use Copilot to track its own decisions, enable reproducibility, and support auditability

By the end, your team will have a fully migrated project—and a blueprint for adopting Copilot at scale without losing trust or control.

Computer Setup:

• Visual Studio Code (preferred)
• GitHub Copilot
• .NET 9+
• SQLite

Timeline
Day 1: Setup, Edit Mode, and Agent Preview
Morning: Laying the Groundwork

• Overview of the project: Dapper-based mailer API
• Why enterprises are moving to lightweight editors like VS Code and Cursor
• Setting up VS Code with Copilot: extensions, modes, and configuration
• Writing effective custom instructions (markdown + JSON)
• Git best practices: .gitignore, branching, and Copilot-generated commits

Afternoon: Edit Mode & Agent Preview

• Step-by-step Entity Framework migration using Edit Mode
• Prompt shaping: how small tweaks yield better results
• Error handling and recovery via conversational feedback
• Agent Mode preview: Running the same EF migration in Agent Mode
• Side-by-side comparison of Edit vs Agent results
• Take-home exercise: create an “audition branch” for your favorite approach

Day 2: Full Agent Mode and AI Self-Audit
Morning: Advanced Agent Mode

• Deep dive into Agent Mode setup and safety
• Using SQL files and markdown to generate models and contexts
• VS Code settings for commit formatting and structural consistency
• Project-wide refactors with controlled automation
• Review flow: understanding, editing, and approving AI output

Afternoon: Auditing, Reproducibility, and Closing the Loop

• Capturing and saving Copilot conversation history
• Tracking refactor decisions through commit messages and chat logs
• Creating an internal “AI audit trail” with reproducible prompts
• Team strategies for integrating Copilot with code reviews and pull requests
• Workshop wrap-up: What worked, what didn’t, and what’s next