Lightrag Server with Raganything

[BireleyX]

Does Raganything work under a Lightrag server instance?
I already install raganthing[all] and libreoffice, but it seems pptx can't be parsed and images in documents
aren't OCRd and mined for information.

Did I miss something or is Raganything no suported when run Lightrag server?

[onestardao]

Hey! You’re definitely not the only one who ran into this — there’s a pretty consistent pain point when running Raganything inside LightRAG server instances, especially when dealing with non-text formats like .pptx or image-heavy documents.

From what I’ve seen, this is usually caused by missing or broken OCR + MIME pipeline integration. Some setups (especially under LightRAG server mode) don’t fully trigger the expected processing chain for binary or embedded content.

Here’s what I’d check:

Is your server mode explicitly enabling OCR handlers? Just installing raganything[all] isn't always enough — some handlers need manual wiring into the document router.

Are pptx, docx, or pdf being properly unpacked before ingestion? (Some pipelines skip embedded images or skip parsing altogether if fallback extractors aren’t mounted.)

Try running your file through the local CLI version of Raganything — if it works there, but not via server, it’s almost certainly a server-path integration issue.

btw I’ve been mapping a bunch of these integration issues (from file parsing to vectorization drift to silent failures downstream), so if you’re debugging RAG stack stuff often, happy to share some resources I’ve compiled. let me know.