pdf-knowledge

Indexing PDFs for semantic search involves parsing documents into a vector and graph knowledge store, enabling fast retrieval, RAG workflows, and relationship extraction across large document collections.

Hybrid search on PDF documents combines vector-based semantic search with graph relationship queries to retrieve highly relevant information. You can execute this using the hybrid_search tool after indexing your files.

Yes, running RAG workflows on PDFs requires the pdf-vector-graph MCP server to be active. This server powers the ask_documents and search_documents tools to extract answers from your indexed knowledge store.

Yes, you can query relationships directly using the query_knowledge_graph tool. Once PDFs are indexed into the graph store, the tool extracts and maps structured connections without manual intervention.

The best way to search large PDF collections is using the ask_documents tool for direct answers or hybrid_search for combined semantic and graph retrieval. Both require indexing PDFs first via the index_pdf tool.

Document indexing in PDF-heavy workflows fails if the required pdf-vector-graph MCP server is not running. Ensure the server is active before calling index_pdf to parse and store documents in the vector and graph knowledge base.