All Tools
O
RAGFreeOpen Source
OPENKB
Compile documents into a living wiki for your LLM
Apache-2.0
ABOUT
Traditional RAG chunks documents, embeds them into a vector database, and re-embeds everything when documents change — a fragile pipeline that loses cross-document context and wastes tokens re-processing stable knowledge. OpenKB instead compiles raw documents (PDFs, Markdown, web pages) into a structured, interlinked wiki of Markdown files using an LLM. The wiki captures concepts, summaries, and relationships once, then stays persistent. Queries use PageIndex's vectorless reasoning-based retrieval that can handle long documents without chunking, making knowledge accumulation both cheaper and more reliable than traditional RAG.
INSTALL
pip install openkbINTEGRATION GUIDE
1. Compile a research library of 100+ PDFs into an interlinked wiki that your LLM can reference without re-embedding
2. Build a continuously updated company knowledge base from Notion exports, Slack archives, and internal docs
3. Create a personal second brain that accumulates and cross-references notes, bookmarks, and articles over time
4. Power a customer-facing Q&A bot with a curated wiki that gets updated by adding files to a directory
5. Replace traditional vector RAG in agentic workflows with a persistent, human-readable knowledge artifact
TAGS
ragknowledge-basellmclidocumentswikiopen-source