GClaude Indexer: Turn a folder of scanned documents into a searchable index

I had a folder of a few hundred scanned documents I was not allowed to upload anywhere. Confidential, mostly images of paper with no text layer, so searching inside them was not possible at all. Opening them one by one was the only way to find anything. Uploading them to a cloud assistant was the obvious answer and also the one I could not use. So I built the step that comes before the assistant instead. It scans the folder, runs OCR only on the pages that actually need it, reads every page, and classifies each one into an item with a type, a date, an author and a summary. The output is four Markdown files you can read on their own or hand to an AI project. Everything runs locally by default. The model server address is fixed in the code rather than read from configuration, so it cannot accidentally point somewhere else. Two honest limits. It is Windows only, because it talks to Windows interfaces directly for hardware detection and the live resource graphs. And the quality score it shows measures how confident the engine declared itself and how completely it filled the fields, not whether it was right. It is free and GPL-3. What I would most like is people running it on hardware that is not mine, since it was built and tested on essentially one machine.