As GitHub Copilot becomes more capable, the difference between average results and exceptional results increasingly comes down to how you ask. Prompting is no longer just a trick, it is a core engineering skill for getting reliable, high-quality outcomes from AI in real-world codebases.

This 30-minute session is designed for developers and technical leads who already use GitHub Copilot and want to push it further. We’ll focus on how to use advanced prompt engineering techniques to make GitHub Copilot work effectively on complex, multi-step engineering tasks, from refactoring legacy systems to designing new features and debugging production issues.

You’ll learn how to:

• Structure prompts so GitHub Copilot understands large, messy codebases
• Use step-by-step and role-based prompting to guide reasoning and design
• Break complex changes into agent-like workflows GitHub Copilot can execute reliably
• Use repository context, constraints, and examples to get predictable output
• Avoid common pitfalls such as over-generalisation, hallucinated APIs, and unsafe code
  
  

We’ll also look at how prompt quality directly affects security, performance, and maintainability, and how to combine strong prompting with code review and GitHub Advanced Security to keep quality high even when AI is doing more of the work.

By the end of the session, you’ll have a practical toolkit for turning GitHub Copilot from a helpful assistant into a reliable collaborator for complex engineering workflows, helping your team move faster without losing control of quality.

This session is ideal for developers, architects, and tech leads working on large or long-lived codebases who want to get more consistent, production-grade results from GitHub Copilot.