Github Copilot - Across Environments

Code Completion with GitHub Copilot

Supported Languages (High Priority):

• Python

• JavaScript

• Java

• TypeScript

• Ruby

• Go

• C#

• C++

Note: Also supports many other languages, but above are strongest.

Auto Suggestions (Ghost Text):

• Real-time suggestions while typing

• Can:

  • Complete a line

  • Generate full code blocks

• Actions:

  • Accept → Tab

  • Ignore → keep typing

• Saves time on:

  • Syntax

  • Logic

  • Repetitive code

Multiple Suggestions Pane:

• Shows alternative solutions

• Navigate:

  • Windows/Linux → Alt + ] / Alt + [

  • Mac → Option + ] / Option + [

• Helps:

  • Compare approaches

  • Choose best implementation quickly

Adaptation to Coding Style:

Copilot learns from your codebase:

• Naming conventions (variables, functions)

• Formatting (indentation, brackets)

• Comment style (inline, block, docstrings)

• Design patterns used in project

• Method implementation patterns

Comments → Key for Better Suggestions

Copilot uses comments via:

• NLP (Natural Language Processing)

• Contextual analysis

Types of Comments Used:

• Inline comments

• Block comments

• Docstrings

• TODO comments

• API documentation

Comment-Driven Code Generation:

1. Function Implementation

   • Comment → full function generated

2. Code Completion

   • More accurate suggestions

3. Variable Naming

   • Suggests meaningful names

4. Algorithm Selection

   • Follows described logic (e.g., bubble sort)

Exam Summary (Important):

• Ghost text = real-time AI suggestions

• Multiple suggestions = compare solutions

• Copilot adapts to your coding style

• Comments significantly improve output quality

• Comments + context = best results

---

GitHub Copilot Chat

What is Copilot Chat:

• AI conversational assistant inside IDE

• Uses natural language to:

  • Generate code

  • Debug

  • Explain code

Main Use Cases:

1. Code Generation

   • Create:

     • Algorithms

     • SQL queries

     • Data structures

   • Example:

     • “Create bubble sort in Python”

2. Debugging

   • Analyze errors and suggest fixes

   • Can explain issues step-by-step

   • Works well with inline chat on selected code

3. Code Explanation

   • Explains complex code

   • Breaks down logic

   • Suggests improvements

Improving Responses (Very Important):

Scope Referencing:

• #file:filename → focus on specific file

• @workspace → entire project

Examples:

• #file:controller.js explain this

• @workspace where is calculate function

Slash Commands (Must Know):

• /doc → add documentation

• /explain → explain code

• /fix → fix errors

• /generate → generate code

• /optimize → improve performance

• /tests → create tests

Tip: Slash commands = better + faster responses

Model Selection:

Standard Models:

• Example: GPT-4o

• Fast, general tasks

• 1 PRU per request

Premium Models:

• Example: o1-preview, o1-mini

• Better for complex reasoning

• 2 PRUs per request

Exam Tip:

• Use premium only for complex problems

Copilot Agents:

• @workspace → full project context

• @workspace /new → create new project

• @terminal → terminal help

• @vscode → IDE help

Feedback System:

• 👍 Helpful response

• 👎 Not helpful

• Improves Copilot over time

Exam Summary (Must Remember):

• Copilot Chat = interactive AI assistant

• Key features:

  • Code generation

  • Debugging

  • Code explanation

• Improve output using:

  • Scope (#file, @workspace)

  • Slash commands

• Agents = context-specific assistance

• PRUs = cost per request (1 vs 2+)

---

GitHub Copilot on GitHub.com

Where Copilot is Used on GitHub.com:

• Repository pages

• Issues & Pull Requests

• Discussions

• Code review

Key Capabilities:

Repository Exploration:

• Explain code, files, and structure

• Generate project summaries

• Create/improve documentation (README, API docs)

Pull Request Assistance:

• Generate PR summaries

• Suggest improvements

• Help resolve merge conflicts

• Suggest documentation updates

Exam Tip:

• PR summaries = 1–2 PRUs

Issue Management:

• Analyze issues

• Suggest solutions

• Create reproduction steps

Code Review & Collaboration:

• Generate review comments

• Detect:

  • Security issues

  • Performance problems

• Improve code quality

Exam Tip:

• Code review = 1–3 PRUs

GitHub Actions Error Explanation:

• Analyze failed workflows

• Identify root cause

• Suggest fixes

• Provide best practices

Agent Tasks:

• Run tasks in background

• Help automate workflows

Exam Summary (Must Remember):

• Copilot works directly on GitHub web interface

• Helps with:

  • Repo understanding

  • PRs

  • Issues

  • Code review

  • Actions errors

• Key benefits:

  • Faster collaboration

  • Reduced manual work

  • Better code quality

• PRUs used for advanced tasks (PRs, reviews, etc.)

---

GitHub Copilot CLI

What is Copilot CLI:

• AI assistant in the terminal (command line)

• Helps:

  • Explain commands

  • Suggest shell commands

  • Interact with files/projects

• Uses GitHub authentication

• Works independently from GitHub CLI

Installation & Launch:

Install:

• brew install copilot-cli

• or install script

Run:

• copilot → interactive mode

• copilot -i "prompt" → one-shot mode

Interactive Mode:

• Chat-like experience in terminal

• Can:

  • Ask questions

  • Generate commands

  • Revise suggestions

• Must trust directory (security check)

Key Features:

Natural Language:

• Ask tasks in plain English

• Example: “find large files and delete them”

Context:

• Use @file to select file context

Important Slash Commands:

• /help → show commands

• /explain → explain command

• /suggest → suggest command

• /revise → improve suggestion

• /feedback → send feedback

• /exit → exit session

• /model → choose AI model

Note:

• Slash commands = required for control/config

Example Workflows:

• Explain command → /explain git reset

• Suggest command → “find .log files”

• Revise → refine previous output

Configuration:

Controlled by:

• Slash commands

• Permissions

Key Controls:

• Trusted directories

• Tool permissions (run/edit commands)

• Path & URL access

Best Practices:

• Use interactive mode for exploration

• Use one-shot mode for quick tasks

• Always review commands before running

• Combine with GitHub CLI (gh)

• Use slash commands for structured control

Exam Summary (Must Know):

• Copilot CLI = AI in terminal

• Two modes:

  • Interactive (copilot)

  • One-shot (copilot -i)

• Slash commands = control system

• Natural language → command generation

• Security = trust directories + permissions