A Practical Guide to
GitHub Copilot
Sina Mostafanejad
ACT-CMS Faculty Fellows Bootcamp
June 2024
1
What is GH Copilot?
``… an AI coding assistant that helps you write code faster and with less effort''
2
Main Features
• Code completion
• Chat
• Pull Request summaries
• Knowledge bases (Copilot Enterprise)
3
Interfaces
• IDE/text editor
• Command line interface (GH CLI)
• Chat interface through GitHub Mobile
• GitHub.com interface (Enterprise subscription)
4
Interfaces
• IDE/text editor
• Command line interface (GH CLI)
• Chat interface through GitHub Mobile
• GitHub.com interface (Enterprise subscription)
5
Getting Started with Copilot
• The following IDEs/text editors are supported
• Microsoft Visual Studio
• Visual Studio Code
• Vim/Neovim
• JetBrains IDEs
• Microsoft Azure Data Studio
https://docs.github.com/en/copilot/using-github-copilot/getting-started-with-github-copilot
6
Getting Started with Copilot
• The following IDEs/text editors are supported
• Microsoft Visual Studio
• Visual Studio Code
• Vim/Neovim
• JetBrains IDEs
• Microsoft Azure Data Studio
https://docs.github.com/en/copilot/using-github-copilot/getting-started-with-github-copilot
7
Supported Languages
• Many programming languages are supported including
• Python,
• JavaScript,
• TypeScript,
• Ruby,
• Go,
• C#
• C/C++
• Copilot can also assist in query generation for databases.
Supported Languages
• Many programming languages are supported including
• Python,
• JavaScript,
• TypeScript,
• Ruby,
• Go,
• C#
• C/C++
• Copilot can also assist in query generation for databases
Your First Suggestion
• Start writing code...
• the suggestions start showing up as you write!
• Accept the suggestions by pressing the tab button
• Partially accept the suggestion by using ctrl + -->
• Hover the mouse over the suggestions to see alternative options
• Use Alt + [ or Alt + ] to switch between alternative suggestions
• Use ctrl + Enter to see a potential list of suggestions in a new panel
Code Completion
• You get suggestions from Copilot as you write code, so...
• Exercise:
• Open the 00_code_completion.py file and instruct Copilot to write you a
simple [e.g., add()] function
Comments Are Valuable!
• Provide as much information as possible
• Offering examples is helpful, especially working with data or strings
• Top-level comments can give context about the overall intended code
• Useful for boilerplate code to get you started
Comments Are Valuable!
• Exercise:
• Open the 01a_comments.py file and use multiple comments to instruct
Copilot to define an add() function, write unit test(s) for it and run the test(s).
• Open the 01b_top_comment_solution.py file and in a top-level comment ask
Copilot to write a complete calculator class with add, subtract, … member
functions. Provide as much detail as possible.
Be Specific, Please!
• All headers, modules and libraries are best to be included/imported manually.
• Be specific about the versions or libraries when asking Copilot
• Exercise:
• Open the 02_specific_versions.py file and instruct Copilot to write a "Hello
World" print statement in Python 2.7 and 3.0 for you!
Context Matters!
• LLMs make inference based on the context
• If you keep relevant code files open in the IDE, Copilot uses their content to make
better suggestions
• Closed files do not contribute to the context.
Chat Interface
• There is a Chat Interface within IDEs that can be used for chatting with Copilot.
• Simply press the Copilot logo on the bottom right bar in the VSCode and
select GitHub Copilot Chat to start, or
• Press Ctrl + Alt + I to open the side chat panel
• Navigate your conversation using up or down arrow buttons on the keyboard
Context Variables (#)
• Use the #editor context variable in the chat interface to provide additional
context from the currently opened files in the VS/VSCode.
• Use #file to attach a file to your instruction/question to provide targeted context
for better outcome
Context Variables (#)
• Exercise:
• Work on 03_context.py to define a Calculator class which
• Implements an addition member function that wraps around the add()
imported from the 00_code_completion.py module.
• After writing the class, call the addition function, and print the result.
• Hint: Keep the 00_code_completion.py open in your editor to provide
context. You may need to rename the file before importing from it.
Context Variables (#)
• Exercise:
• Using context variables (#), provide additional context for the Calculator class
and ask Copilot in the chat how a subtract function can be added to
the 00_code_completion.py module or the Calculator class
Naming Conventions
• Give your functions and variables meaningful names
• Meaningful names create better codes
• Meaningful names generate better context and therefore, better suggestions
from Copilot
• Exercise:
• Open the 04_naming_convention.py file and define a function with a random
name [e.g., asdfjkh23m()] and see what Copilot suggest for its body.
Examples Help!
• As humans learn the new concepts better with specific examples, AI algorithms
can do too.
• In your instructions and comments, try to provide specific examples (e.g., of the
expected output, return values etc.)
• Exercise:
• Open the 05_examples.py file and instruct Copilot to write a function that
takes two arrays of integers as input and returns the sum of the two arrays.
Inline Chat
• Chat can be done in an inline fashion
• Press ctrl + I to see a pop-up chat bar.
• Useful for quick fixes with code diffs and documentation
• Highlighting the relevant code narrows down context and helps with
the suggestions
• Look for Magic Sparkles to get help from Copilot Inline chat
Inline Chat
• Exercise:
• Open the 05_examples_solution.py file and instruct Copilot through inline
chat to write NumPy/Google style docstring for your function(s)
Slash Commands (/)
• The Slash Commands are designed for common tasks
• /doc ---> Add documentations for objects
• /explain ---> Explain the highlighted code
• /fix ---> Provide a potential fix for the highlighted problematic code
• /generate ---> Generate code as instructed
• /help ---> Get help on Copilot Chat
• /optimize ---> Analyze and enhance efficiency of the highlighted code
• /simplify ---> Simplify the highlighted code
• /tests ---> Write unit test for the highlighted code
• /clear ---> Clear the chat
Copilot Agents (@)
• Agents can help with a large variety of tasks providing context on their own.
• Instead of providing context in our prompts, we can ask Copilot to build the
context on its own.
• Currently there are 3 agents in Copilot:
• @workspace ---> Context from workspace
• @vscode ---> Questions related to VSCode and its structures
• @terminal ---> Chat pertinent to the terminal commands
Copilot Agents (@)
• @workspace builds the context from our workspace and can be used for:
• Looking for files, searching for modules, class or function definitions etc.
• Adding new functionalities
• Fixing the code issues and errors
• Suggestions for refactoring/restructuring the code
• Exercise:
• Close all files in the editor and open the Copilot chat interface. Use the @workspace agent and
ask where is the add() function defined?
References
• https://docs.github.com/en/copilot/using-github-copilot/using-github-copilot-
code-suggestions-in-your-editor
• https://github.blog/2024-03-25-how-to-use-github-copilot-in-your-ide-tips-tricks-
and-best-practices/
• https://github.blog/2023-05-17-inside-github-working-with-the-llms-behind-
github-copilot/
• https://github.blog/2023-05-17-how-github-copilot-is-getting-better-at-
understanding-your-code/
• https://dev.to/github/a-beginners-guide-to-prompt-engineering-with-github-
copilot-3ibp
• https://medium.com/@yar.dobroskok/github-copilot-workspace-new-
development-experience-d69857fbd067
