OpenKB is an open-source knowledge base system from VectifyAI, the team behind PageIndex, that compiles raw documents into a structured, interlinked, wiki-style knowledge base an LLM can actually reason over. The command-line tool has gathered roughly 4,800 stars on GitHub since its 2026 debut and offers an alternative to classic chunk-and-embed RAG for long, complex documents.
Instead of shredding documents into anonymous chunks, OpenKB reads them and builds persistent wiki pages in the Google Open Knowledge Format (OKF): concepts get their own pages, cross-linked to each other, and the whole base stays human-readable and browsable. Under the hood it uses PageIndex tree indexing to navigate long documents structurally, so retrieval follows the document’s own organization. Agents and chat interfaces can then query the compiled knowledge base rather than re-reading sources from scratch. Because the output is plain, version-controllable wiki pages rather than embeddings in a database, teams can review, correct and diff their knowledge base like code, which makes the approach attractive for high-stakes documentation where provenance matters.
Free and open source under the Apache 2.0 license. You run the CLI yourself and pay only for whatever LLM API you connect to it.
Developers and research teams building document intelligence, internal wikis and RAG systems who want structured, inspectable knowledge instead of an opaque vector store.