ReformedPapr crawls a URL, downloads every PDF it finds, converts them to searchable text, and lets you chat with the whole collection — answers come with citations back to the source document.
Invite only — powered by Ollama running locally
Point it at a URL and walk away. ReformedPapr handles the crawl, the download, and the ingestion — you just ask questions.
Add a URL — ReformedPapr spiders the entire site, discovers every PDF linked from it, and queues them all for download. No manual hunting required.
PDFs are converted to plain text in under a second using pdftotext. No OCR delays, no cloud uploads — runs entirely on your machine.
Every document is chunked and indexed in a SQLite FTS5 full-text index. Searches are instant across thousands of pages, even on a laptop.
Ask questions in plain English. The most relevant chunks are retrieved, passed to a local LLM (Ollama), and an answer is synthesized with inline citations.
Every answer links back to the exact document — with a download link to the original PDF and a link back to the source URL it was crawled from.
No data leaves your machine. PDFs stay on disk, the LLM runs via Ollama locally, and the index lives in SQLite. No API key required.
Paste a URL to crawl (discovers all PDFs on the site) or a direct PDF link. ReformedPapr handles both — crawl a whole library or grab a single document.
PDFs are downloaded and stored locally. Each one is converted to searchable text in under a second — no cloud services, no waiting.
Text is split into ~500-token chunks and indexed in a full-text search database. The index is ready to search the moment a document finishes processing.
Ask any question. The top matching chunks are retrieved and sent to the LLM, which generates a grounded answer with links back to the source PDFs.
Stop ctrl-F-ing through PDFs. Index the whole collection and ask it anything — answers back in seconds, citations included.