Class VisionExtractTool

All Implemented Interfaces:
GuardedExecution, Identifiable, Job<Extraction>, ModelDependent, ModelDependentTool<FileRef, Extraction>, Tool<FileRef, Extraction>

public class VisionExtractTool extends AbstractModelDependentTool<FileRef, Extraction>
The tier for what the code cannot read: an image, or a PDF. The file is shipped whole to a model - the image as an image, the PDF as a document - and comes back as text; nothing is rendered locally. The tool declares no model: it declares a grade and what it sends (images or documents, by the file's bytes), so the picker resolves an entry that accepts it, whichever provider serves it.
Since:
0.1 (2026-09-14)
Author:
Andrey Santrosyan