Duration: 45 minutes (15 min demo + 30 min hands-on)
Format: Demo + Hands-on
By the end of this module, you will be able to:
- Use Copilot Chat on GitHub.com to query repositories, explain code, and explore codebases
- Generate pull request summaries with Copilot
- Use Copilot code review to get AI-powered review feedback on pull requests
- Leverage Copilot from the search bar and dashboard for repository-scoped questions
Copilot Chat is available across GitHub.com:
| Access Point | How to Open | Best For |
|---|---|---|
| Copilot icon (title bar) | Click the Copilot icon next to the search bar | General questions, cross-repo queries |
| Search bar | Type a question in a repo's search bar → "Ask Copilot" | Repository-scoped questions |
| Dashboard | Prompt box on your github.com dashboard | Quick questions on login |
| PR / Issue context | Open Copilot while viewing a PR or issue | Context-aware code questions |
Copilot dynamically selects skills based on your question. Key skills include:
| Skill | What It Does | Example Prompt |
|---|---|---|
| Code search | Searches across your repository code | Where is the authentication middleware defined? |
| Bing web search | Answers with current web information | @github #web What is the latest LTS version of Node.js? |
| Path search | Finds files by path pattern | Find all configuration files in this repo |
| Issue/PR search | Searches issues and pull requests | What are the open P1 bugs assigned to my team? |
Enterprise tip: Use
What skills are available?to see the full list of skills Copilot can use.
- Subthreads: Branch a conversation by editing or retrying any question
- Model switching: Regenerate a response with a different AI model using the retry icon
- File generation: Copilot can generate complete files that you can view, edit, and download
- Sharing: Share conversations with teammates via a link (preview feature)
Copilot can automatically generate a summary of the changes in a pull request.
- Create or open a pull request
- In the PR description field, click the Copilot icon (or the "Generate summary" button)
- Copilot analyzes the diff and generates:
- Overview of what changed and why
- File-by-file breakdown with descriptions of each change
- Review guidance highlighting what a reviewer should focus on
| Benefit | Impact |
|---|---|
| Faster reviews | Reviewers get context before reading code |
| Consistent quality | Every PR has a meaningful description |
| Knowledge transfer | New team members understand changes without asking |
| Compliance | Auditable descriptions of all code changes |
Copilot can review pull request code and provide AI-generated feedback.
- Automated review comments: Copilot identifies potential issues, bugs, and improvements
- Custom instructions: Use
.github/copilot-instructions.mdto customize review behavior - One-click request: Request a Copilot review just like requesting a human reviewer
- Inline suggestions: Copilot provides code suggestions directly in the review
| Scenario | How Copilot Helps |
|---|---|
| Security review | Flags potential vulnerabilities, injection risks, credential exposure |
| Code quality | Identifies duplicated code, complex methods, missing error handling |
| Standards compliance | Checks against team-defined coding standards via custom instructions |
| Performance | Highlights N+1 queries, unnecessary allocations, missing caching |
The Copilot coding agent is an autonomous AI agent that works directly on GitHub.com:
- Assign an issue to Copilot → Copilot creates a branch and starts working
- Copilot creates a pull request → You review the proposed changes
- Iterate with comments → Leave feedback, Copilot revises
Enterprise relevance: The coding agent respects your repository's custom instructions (
.github/copilot-instructions.md) and runs CI/CD pipelines to validate its changes.
- Navigate to a repository on GitHub.com (use your organization's repo or a public repo)
- Click the search bar at the top of the repository page
- Ask the following questions and observe how Copilot uses different skills:
What does this repository do?
Where is the main entry point of the application?
What testing frameworks are used in this project?
Show me all API endpoint definitions in this codebase
- Try asking with a web search:
@github #web What are the current best practices for the framework used in this repo?
- Create a new branch and make a small code change (e.g., add a utility function, fix a comment, or update a configuration)
- Create a pull request from the branch
- In the PR description, click the Copilot summary button
- Review the generated summary:
- Is the "what changed" section accurate?
- Does the file-by-file breakdown match your changes?
- Would a reviewer find this summary useful?
- Edit the summary to add any context Copilot missed
- On the same pull request, go to the Reviewers section
- Request a review from Copilot
- Wait for Copilot to analyze the PR and provide comments
- Review Copilot's feedback:
- Are the suggestions relevant?
- Did it catch any issues you missed?
- How does it compare to a human review?
- If Copilot provided inline suggestions, examine whether they're appropriate
- Navigate to your GitHub dashboard (github.com)
- Use the Copilot prompt box to ask:
What are my most recent pull requests that need attention?
Summarize the open issues in [repo-name]
- Copilot on GitHub.com is context-aware: It understands the repository, PR, or issue you're viewing
- PR summaries save review time: Auto-generated summaries help reviewers focus on what matters
- AI code review complements human review: Copilot catches patterns humans might miss, but always verify
- Skills are dynamically selected: Copilot chooses the right tool (code search, web search, etc.) based on your question
- Copilot coding agent enables autonomous work: Assign issues directly to Copilot for background implementation