Reading the Prompt Preview — Your Agent's Final Instructions

Understand the live preview panel and quality score.

3 min read

What is the preview panel?

Look at the right side of your screen. That panel is the preview — and it's the most important thing on the page. This is the "prompt" — the finished product that all your nodes combine to create.

Everything you've been building — the role, the rules, the examples — gets stitched together into one set of instructions. The preview shows you exactly what your AI agent will actually read when it starts working.

How it's organized

Each section in the preview matches a node on your canvas. And the colors match too — blue text comes from your Role node, red from your Rules node, and so on. This makes it easy to trace any part of the final prompt back to the node it came from.

The order of sections in the preview follows the connection lines on your canvas. If you rearrange nodes later, the preview order changes to match.

It updates in real time

Try this: click any node on the canvas, change something — even just a single word — and watch the preview update instantly. There's no save button, no refresh needed. What you see is EXACTLY what your agent will read.

This is your feedback loop. Edit on the left, check the result on the right. If something doesn't look right in the preview, go back to the node and fix it.

Clicking a Role node → changing the agent name → preview panel updating instantly on the right to reflect the new name

Video • 15 seconds

Understanding the Quality Score

At the bottom of the preview panel, you'll see a colored bar with a number. That's your Quality Score. It tells you how complete and well-structured your prompt is — at a glance.

Below 40 (Red) — Missing critical sections. Your prompt might work, but it'll be unreliable. The agent is guessing too much on its own.
40–70 (Yellow) — Decent foundation. You've got the basics covered, but adding examples or knowledge sections will make a noticeable difference.
Above 70 (Green) — Solid. Your agent has what it needs to do a good job. You can always improve, but you're in great shape.

Below the score, you'll see specific suggestions — things like "Add 2 example conversations to improve your score." These are tailored to what's missing from your prompt, so they're worth reading.

Don't chase 100

A score of 75 with great, specific examples is better than a score of 100 filled with generic filler. The score is a guide, not a grade. Focus on making your content genuinely useful to the agent, and the score will take care of itself.

Full walkthrough: editing multiple nodes → watching preview update after each edit → scrolling to quality score bar → reading improvement suggestions

Video • 30 seconds

You just learned how to read the preview panel — the live view of your agent's actual instructions. You know that sections match your nodes, colors help you trace content back to its source, and everything updates in real time. You also understand the Quality Score: aim for green, follow the suggestions, but don't stress about a perfect 100.