What does "auto-read charts on upload" do?

When you attach a chart image to a trade, it runs a vision read on it straight away and fills in the Market Context fields you left blank. It never overwrites anything you typed, and it runs once per trade.

This is the feature that turns a screenshot into structured data, and the rules it follows matter more than what it does.

What it fills: the blank Market Context fields — CVD bias, VWAP position, order flow — plus stop-loss and take-profit prices if they are visibly marked on the chart, and any of your Key Levels the price interacted with.

What it will not do. It only fills values that are empty. Anything you typed stays. And it is written to prefer silence over a guess: if it cannot identify a level, it returns nothing rather than naming the one that would probably be there. A thin read on a busy chart usually means the chart was unreadable at capture resolution, not that the feature is broken.

It runs once per trade. One paid vision pass, then it stops. To run it again — after attaching a better image, or after filling in your context profile — use the Read chart button on the trade.

Turning it off makes the read explicit: nothing happens on upload, and Read chart is how you ask. That is the right setting if you attach a lot of images and only want a few of them read, because this is the most expensive routine call the app makes.

Two things make it much better, and both are free: filling in your context profile so the model knows your colour scheme, and capturing the whole screen rather than one chart so it has every timeframe to read from. See why capture a whole screen.