Practical overview
When it fits
Use Openkb when source material needs to be compiled into a persistent, interconnected structure for later retrieval, rather than handled as isolated documents for each query.
Who it is for
- Developers building self-hosted LLM or RAG systems
- Teams organizing long internal documents
- Users needing multimodal retrieval across text, figures, tables and images
- People creating entity-centered or wiki-style knowledge pages
Documented uses
- Compile raw documents into a persistent knowledge base
- Retrieve information from long and complex documents
- Ask questions or chat over a compiled wiki
- Maintain linked pages for people, organizations, places and products
- Generate agent skills, knowledge graphs and HTML slide decks
Requirements & limits
The project is documented as open source under the Apache-2.0 license. It is available as a Python package, CLI and self-hosted web interface. The bundled web interface runs locally and is unauthenticated by default; a bearer token can be configured for server exposure.
- The workflow is CLI-first, although a bundled local web Workbench is documented.
- Retrieval is described as vectorless and reasoning-based, using PageIndex tree indexing for long documents.
- The system compiles summaries, concept pages, entity pages and cross-links into a persistent wiki.
- Supported source types include documents, spreadsheets, text and URLs, with native handling of figures, tables and images.
- The web interface requires the web extra and serves the API and Workbench locally.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install and initialize a local Openkb knowledge base
Set up Openkb, configure an LLM provider and create a knowledge-base directory.
Before you start
- Python package installation capability
- An LLM provider configuration
- An LLM API key, unless using a supported OAuth device-flow provider
Steps
Install Openkb with pip.
Create and enter a directory for the knowledge base.
Run the initialization command and set the model during initialization or in the configuration file.
Create a .env file containing the LLM API key when required.
Expected outcome
A configured, initialized local knowledge-base directory ready for documents.
Guide limitations
- This guide covers setup only; it does not verify a completed query or compilation.
- Provider authentication requirements vary according to the documented provider flow.
Add documents and query the knowledge base
Add documented source material and ask a question through the Openkb CLI.
Before you start
- An initialized Openkb knowledge base
- A configured LLM provider
- A supported document, directory or URL
Steps
Add a file, directory or URL with the document-add command.
Ask a question with the query command.
Alternatively, start an interactive chat session.
Expected outcome
A query or interactive chat is initiated over the knowledge base.
Guide limitations
- The evidence does not document output quality, processing time or successful execution in a particular environment.
- The available retrieval result depends on the configured LLM and supplied source material.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 1 October 2026
- Original website title
- VectifyAI/OpenKB · Official homepage title · retained in its source language