Image-To-Code

Scribe Studio can read chart screenshots and attempt to recreate the visible indicator behavior in Pine Script. Use it when you can see what an indicator does but cannot describe the math.

What To Include In The Image

A good reference screenshot shows:

• The full chart area, not just a thumbnail.
• The indicator output you want to recreate — plots, arrows, fills, labels, zones.
• Enough candles to see the behavior across more than one signal.
• Any labels, lines, zones, or colors you care about.
• The symbol and timeframe, when relevant to the logic.

You can attach up to 5 images per message, with each image up to 5 MB. Clear images produce better results than blurry or cluttered ones.

Add Text With The Image

Always describe what you want in text alongside the image. Screenshots tell Scribe what you see; the text tells Scribe what to build. A typical Image-to-Code message looks like this:

Analyze this screenshot and create an indicator that recreates
the visible signal arrows and trend coloring.

Focus on the green arrows below candles for bullish signals,
the red arrows above candles for bearish signals, and the bar
coloring that follows the active signal direction.

Add alert conditions for both sides.

If the exact logic cannot be inferred from the image, approximate
it with common technical rules and explain the assumptions.

Be explicit about which visual elements matter — Scribe will treat the elements you name as the targets to reproduce, and treat the rest as decoration. You do not need to annotate the screenshot itself; calling out the elements in the text is enough.

What Scribe Can And Cannot Infer

Scribe can usually identify visual patterns, labels, zones, colors, and approximate behavior — the shape of an indicator. It is good at "two EMAs with a fill between them", "RSI divergence labels", "support and resistance boxes".