█

AI/TLDR

OpenKB

Compiles a document pile into an interlinked wiki once, instead of re-deriving it on every query

GitHub★ 4.5kWebsiteopenkb.ai
RAG Frameworks & PlatformsOpen source
Language
Python
License
Apache-2.0
$pip install openkb

Overview

OpenKB is an open-source CLI from VectifyAI that turns a folder of raw documents into a structured, interlinked wiki-style knowledge base. The argument it makes against conventional RAG is about where the work happens: traditional retrieval rediscovers the same knowledge from scratch on every query and nothing accumulates. OpenKB compiles the corpus once into summaries, concept pages, entity pages and cross-references, then keeps that wiki in sync as documents change.

Retrieval is handled by PageIndex, the same team's vectorless, reasoning-based index, which walks a tree structure over long documents rather than matching embeddings — so there is no vector database to run. It ingests PDF, Word, Markdown, PowerPoint, HTML, Excel, CSV, plain text and URLs, and handles figures, tables and images rather than text alone.

On top of the wiki sit generators: one-off `query`, multi-turn `chat` with resumable sessions, an interactive knowledge graph, single-file HTML decks, and a Skill Factory that distils redistributable agent skills out of the knowledge base. The wiki itself is plain Markdown with cross-links, so it opens in Obsidian, and its pages follow Google's Open Knowledge Format. Models are configured through LiteLLM, so OpenAI, Anthropic, Gemini and others all work.

What it does

  • Compiles documents into a persistent wiki of summaries, concept pages and auto-extracted entity pages, kept in sync as sources change
  • Vectorless, reasoning-based retrieval via PageIndex tree indexing — no vector database to deploy
  • Broad ingestion: PDF, Word, Markdown, PowerPoint, HTML, Excel, CSV, plain text and URLs
  • Native multi-modality — retrieves and reasons over figures, tables and images, not just text
  • Skill Factory distils redistributable agent skills from the compiled wiki
  • Output is plain `.md` files with cross-links, Obsidian-compatible and following the Google Open Knowledge Format
  • Bundled Knowledge Workbench web UI for browsing, uploading and streaming queries in the browser
  • Any LiteLLM-supported provider, set at `openkb init` or in `.openkb/config.yaml`

Getting started

OpenKB installs as a Python package and runs as a CLI inside a directory that holds your knowledge base. You need an LLM API key for whichever provider you configure.

Install the CLI

Install from PyPI. You can also install the latest from GitHub, or clone and `pip install -e .` for development.

bashbash
pip install openkb

Initialize a knowledge base and add documents

Create a directory, initialize it, then add files, whole directories, or URLs. `init` is also where you pick the model.

bashbash
mkdir my-kb && cd my-kb
openkb init

openkb add paper.pdf
openkb add ~/papers/                            # a whole directory
openkb add https://arxiv.org/pdf/2509.11420     # or fetch from a URL

Point it at a model

Models use the LiteLLM `provider/model` format (for example `anthropic/claude-sonnet-4-6`); OpenAI models can omit the prefix. Set the model during `openkb init` or in `.openkb/config.yaml`, then put the key in a `.env` file. Subscription providers that use an OAuth device flow need no API key.

bashbash
LLM_API_KEY=your_llm_api_key

Query or chat over the wiki

Ask one-off questions, or open an interactive session that persists so you can resume it later.

bashbash
openkb query "What are the main findings?"
openkb chat

Turn the wiki into other outputs

The generators reuse the compiled knowledge base rather than re-reading the sources.

bashbash
openkb skill new my-expert "Reason like an expert on <your-topic>"   # a portable agent skill
openkb visualize                                                     # an interactive knowledge graph
openkb deck new my-deck "An intro deck on <your-topic>"              # a single-file HTML deck

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

When to use it

  • Building a durable, browsable wiki over a research corpus so synthesis reflects everything ingested rather than the top-k chunks of one query
  • Working with long, structured documents — reports, filings, manuals — where chunk-and-embed retrieval loses the document's hierarchy
  • Querying a document set whose meaning lives in figures and tables as much as in prose
  • Distilling a body of internal documentation into a portable agent skill other tools can load

How OpenKB compares

OpenKB alongside other open-source rag frameworks & platforms tools AI/TLDR tracks, ranked by GitHub stars.

ToolStarsWhat it does
Dify★ 156kAn open-source platform with a visual workflow builder for creating LLM and RAG applications without writing much code.
graphify★ 119kTurns a folder of code, docs, PDFs and images into a local knowledge graph with tree-sitter AST parsing and Leiden communities — queryable by agents over MCP, no vector store.
RAGFlow★ 90.9kA RAG engine built around deep document understanding that turns complex files into a grounded, citation-backed question-answering layer.
Pathway★ 62.3kA Python framework with a Rust streaming engine that keeps ETL, real-time analytics and RAG pipelines continuously up to date as source data changes.
Context7★ 62.2kContext7 pulls current, version-specific documentation and code examples for any library and feeds them into your LLM, available as a CLI skill or an MCP server.
LightRAG★ 39.7kA graph-based RAG system that builds an entity-and-relationship knowledge graph for fast retrieval and easy incremental updates.
Quivr★ 39.5kQuivr is an open-source RAG framework that ingests your documents and answers questions about them, working with any LLM and any file type.
OpenKB★ 4.5kCompiles a document pile into an interlinked wiki once, instead of re-deriving it on every query