Bonus: Free Exploration
You've experienced the full workflow: content changes, design updates, scaling with bulk import, and page import. Now it's your turn to explore. The Experience Modernization Agent is a full AI agent, not just a migration tool. It can respond to virtually any development request.
Use this time to experiment. Ask it anything, prompt it in any language, push its boundaries. This is where you learn what's possible.
1 — Try your own language
Prompt in your mother tongue. (or try a different language)
- If English is not your first language, try prompting in your native language. The agent understands and responds fluently in any language.
2 — Get creative
Try fun prompts.
- Ask the agent to write in pirate slang, explain what it did as a haiku, or suggest a color palette inspired by a season. The agent is creative and playful.
3 — Ask questions
Ask the agent anything about Edge Delivery Services.
- The agent is deeply knowledgeable. Try questions like:
- "What is the best way to implement a carousel in EDS?"
- "How do section metadata and block variants work?"
- "What are best practices for content modeling?"
- If you hit a snag, ask the agent. It's designed to be a guide, not just an executor.
4 — Give instructions
Shape how the agent works.
- You can direct the agent's behavior with explicit instructions:
- "Always use CSS custom properties for colors."
- "Do not modify the header block."
- "Document every new block you create in PROJECT.md."
- In a real project, you'd capture these in an AGENTS.md file so the agent remembers them across sessions.
What just happened
You experienced the agent as what it truly is: a versatile AI development partner, not a narrow migration tool. It can import content, modify designs, answer questions, follow project-specific instructions, and work in any language. Whether you need a quick tweak or a full site migration, the same agent handles it.
Tips for working with the agent
These tips will help you get the best results in this lab and beyond. We've split them into two groups: essential tips you should use right now, and advanced tips for deeper work later.
Essential tips (use these now)
1. Describe the goal, not the implementation
The agent works best when you explain what you want to achieve and why.
❌ Instead of: "Add a CSS rule: .hero { background-color: #1a1a1a; }"
✓ Say: "Make the hero section background darker. The white text is hard to read."
The agent understands intent and figures out the best implementation.
2. One thing at a time
Break complex work into smaller prompts. Each prompt focuses on one change. This lets you review before moving on.
❌ Instead of: "Import the page, fix the header, restyle buttons, and add a footer"
✓ Do: Import first, verify, then restyle, verify, then add the footer.
3. Iterate, don't restart
If something doesn't look right, ask the agent to fix that specific issue. Iterative refinement is faster than starting over.
❌ Instead of: Clearing the chat and re-prompting from scratch
✓ Say: "The hero heading is too small, make it larger" or "The button should be darker green."
4. Stay in control
If the agent is heading the wrong direction, interrupt it immediately.
- Ask: "Stop. What are you about to do?"
- If it already made changes you don't want, tell it to revert and clarify what you actually need.
- For complex tasks, use the Plan feature in the Console: the agent outlines its approach before executing, so you can review and adjust.
Advanced tips (for deeper work later)
1. Give the agent project context
The agent infers your project's naming conventions and patterns from the code it reads. Before larger tasks, ask it to review reference files:
- "Look at how blocks are named in the blocks/ folder."
- "Review hero.css to see how CSS is structured in this project."
When the agent understands your patterns, its code changes are cleaner and match your existing style without extra cleanup.
2. Capture instructions in AGENTS.md
In real projects, create an AGENTS.md file with project-specific rules:
- "Always use CSS custom properties for colors"
- "Do not modify the header block"
- "Document all new blocks in PROJECT.md"
The agent reads this file at the start of each session and follows your rules automatically.
3. Maintain a PROJECT.md file
Ask the agent to document what it creates:
- "Document all the blocks and CSS tokens you've created in PROJECT.md"
- "Summarize everything you've done and save it to PROJECT.md"
This file becomes a running record: blocks created, design decisions, CSS tokens, templates. If you restart a session, the agent reads AGENTS.md and PROJECT.md and has full context to continue.
4. Handle interruptions
- If the agent stops unexpectedly, type "continue" to resume.
- If the agent appears stuck for 5–10 minutes with no progress, interrupt and type "continue" to restart from where it left off.
Success You've explored the full breadth of the Experience Modernization Agent: content adaptation, design changes, bulk import, page import, and free exploration. You're ready for the conclusion.