Iterating on Components

You rarely get a perfect result on the first prompt. The agent is designed for iterative refinement — send follow-up prompts to adjust, extend, or rework your component.

Follow-Up Prompts

After the agent generates a component, you can send additional prompts to refine it:

• “Move the chart below the metrics instead of beside them”
• “Change the header text color to blue and make it larger”
• “Add a loading spinner while the data is being fetched”
• “Replace the table with a description list for fewer fields”

The agent understands that follow-up prompts refer to the current component and applies changes incrementally.

Conversation History

The editor maintains a conversation log showing all prompts and agent responses within your session. You can scroll through the history to review:

• Your prompts — what you asked for
• Agent status messages — processing indicators
• Agent thinking — the agent’s reasoning about your request
• Completed results — the generated or updated code

This history gives the agent context about what has already been built, so follow-up prompts can be brief and focused.

Version History

The editor tracks versions of your component. Each time the agent generates or updates code, a new version is created. You can review previous versions to compare changes over time.

Tips for Effective Iteration

• Be incremental — change one thing at a time rather than rewriting the entire prompt
• Be specific about what to change — “make the button green” is better than “improve the styling”
• Reference existing elements — “move the save button to the top right” tells the agent exactly which element to modify
• Ask for alternatives — “show me a different layout for the same data” lets the agent explore other approaches
• Start over when needed — if the component has drifted too far from your vision, write a fresh detailed prompt instead of layering more changes