MinerU Document Explorer

Agent-native knowledge engine — search, deep-read, and build knowledge bases from Markdown, PDF, DOCX, and PPTX. Supports hybrid search with BM25, vector search, and LLM reranking. Optional external data files include LLM models auto-downloaded on first use (~300MB to ~1.1GB). Python packages are optionally required for PDF, DOCX, and PPTX processing.