Skip to content

01-setup.md

Latest commit

 

History

History
142 lines (91 loc) · 3.83 KB

01-setup.md

File metadata and controls

142 lines (91 loc) · 3.83 KB

Part 1: Setup & Context Engineering

📚 Lab Guide • ← Overview

⏱️ Time: ~15 minutes

In this section, you'll set up your development environment and teach GitHub Copilot about your codebase.

🔧 Initial Setup

Step 1: Create Your Repository

  1. Open github.com/copilot-dev-days/agent-lab-java
  2. Click Use this template → Create a new repository
    • Name: my-soc-ops-java
    • Visibility: Public
  3. ✅ Your own Soc Ops repo is ready!

Step 2: Enable GitHub Pages

  1. Go to your repo's Settings → Pages
  2. Under "Build and deployment", select GitHub Actions
  3. ✅ Every commit will now publish to: https://{username}.github.io/{repo-name}

Step 3: Choose How You'll Work (Local Clone or Codespace)

After creating your own repository, choose one:

Option A: Clone Locally in VS Code

  1. Open VS Code
  2. Run command: Git: Clone → Clone from GitHub
  3. Select your new repository (not the template repo)
  4. When prompted, install recommended extensions

Option B: Create a Codespace for Your Repository

  1. Open your new repository on GitHub
  2. Click Code → Codespaces
  3. Click Create codespace on main
  4. Wait for setup to finish

Step 4: Run the Setup Agent

In the Chat panel:

/setup

The agent will:

  • Detect your environment
  • Install any missing dependencies
  • Start the development server

✅ Success: App is running in your browser!

📚 Understanding Context Engineering

Context engineering is how you teach AI about your specific codebase. This makes Copilot's suggestions more accurate and relevant.

Task 1: Generate Workspace Instructions

Instructions guide all agentic interactions, making them efficient and reliable.

Steps:

  1. Run command: Chat: Generate Workspace Instructions File
  2. Wait for the agent to analyze your codebase
  3. Review the generated instructions (probably too detailed!)
  4. Follow up with: 
    Compress down by half and add a mandatory development checklist 
(lint, build, test) to the top
  5. Commit the instructions file

✅ Result: All future requests have a basic map of your workspace.

Task 2: Background Agents for Parallel Work

Background agents run in isolated git worktrees — perfect for tasks that don't need handholding.

Steps:

  1. Click + in Chat → New background agent
  2. Enter: 
    Add linting rules for unused imports and code style; fix any errors
  3. Let it run, then Review and Apply the changes
  4. Right-click the session to delete it when done

Try a Cloud Agent too:

  1. Click + → New cloud agent
  2. Enter: 
    Make the README more engaging as a landing page to the project

✅ Result: Linting rules added, errors fixed, README improved — all merged back to main!

Task 3: Explore Existing Instructions

Your repo comes with pre-configured instructions that help the AI understand the project.

CSS Utilities Instructions

📄 See .github/instructions/css-utilities.instructions.md

These document the custom Tailwind-like CSS classes available in this Spring Boot project.

💡 Optional: Delete the main text and re-run the prompt to see how it generates

Frontend Design Instructions

📄 See .github/instructions/frontend-design.instructions.md

The "no purple gradients" instructions challenge the agent to think like a designer.

💡 Think about: What other AI biases could you challenge and nudge?

✅ Part 1 Complete!

You've learned how to:

  • Set up your development environment
  • Generate and refine workspace instructions
  • Use background and cloud agents for parallel work
  • Understand existing instruction files