How to have an AI agent read charts, diagrams, and scanned pages Leave feedback

Text extraction sees characters. It does not see a bar chart, a signature stamp, a diagram, or a scanned page that is only a picture. Rendering the page and giving the image to a vision-capable model covers exactly that gap.

Note

The commands and config snippets on this page are for the .NET build of the server — the only platform available today. Installation and client setup: MCP server for .NET. Other platforms will expose the same tools with their own launch command; everything else on this page applies unchanged.

On page 7 of the annual report there is a revenue chart. What were the figures for each quarter?

The agent renders page 7 with render_page and reads the values from the image.

• Charts and graphs — values, trends, labels, and legends.
• Scanned pages — a PDF made from a scanner has no text layer; the image is the only way in.
• Stamps, signatures, and handwriting — whether a page is signed, and what the stamp says.
• Layout questions — “is the logo in the header?”, “does this slide overflow?”.

A model reading an image can misread a small number or a faint scan. For figures that matter, ask the agent to say how sure it is, and check against the source when the answer goes into a decision. Rendering does not turn a scan into verified text.

For a long document, find the page before rendering:

How many pages does the report have? Render the contents page and tell me where the revenue chart is.

Two small renders instead of fifty.

In evaluation mode the watermark sits on top of the page, which can hide exactly the detail you asked about. Check get_license_status before relying on what the agent reads.