RAG Frameworks · VectifyAI

OpenKB

OpenKB is an open-source Python CLI tool that compiles raw documents into a structured, wiki-style knowledge base using LLMs and vectorless retrieval. It organizes documents into summaries, concept pages, and cross-referenced entities, then lets you query, chat, and generate agent skills from the compiled knowledge.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/VectifyAI/OpenKB

2.9k

GitHub stars

320

Forks

Python

Primary language

Apache-2.0

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	VectifyAI/OpenKB
Owner	VectifyAI
Primary language	Python
License	Apache-2.0 — OSI-approved
Stars	2.9k
Forks	320
Open issues	24
Latest release	v0.4.3 (2026-07-02)
Last updated	2026-07-08
Source	https://github.com/VectifyAI/OpenKB

What OpenKB is

OpenKB integrates markitdown (document parsing), PageIndex (hierarchical tree indexing for long PDFs), and LiteLLM (multi-provider LLM abstraction) to build a persistent, compiled wiki in plain Markdown with cross-links. It supports multi-modal retrieval (text, tables, images) and follows Google OKF specification for knowledge representation.

Quickstart

Get the OpenKB source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/VectifyAI/OpenKB.git2cd OpenKB3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Research and Paper Synthesis

Accumulate insights from multiple papers, PDFs, and research URLs into a structured wiki with auto-extracted concepts and entity pages. Enables reasoning-based retrieval across long documents without vector databases.

Internal Knowledge Compilation

Compile internal docs, runbooks, design specs, and wikis into a persistent, interlinked knowledge base that grows with your org. Knowledge compounds; contradictions are flagged automatically.

Agent Skill Distillation

Extract redistributable agent skills from your wiki using the Skill Factory, turning accumulated knowledge into portable, versioned capabilities for downstream agents or automation pipelines.

Implementation considerations

LLM API dependency: All knowledge compilation and query reasoning rely on external LLM calls (OpenAI, Claude, etc.); budget API costs and set up `.env` LLM_API_KEY before init.
Document format support breadth: PDF, Markdown, Word, PowerPoint, HTML, Excel, CSV, text, and URLs supported; test complex layouts (multi-column, embedded media) with your corpus first.
Long document handling via PageIndex: PDFs ≥20 pages auto-route through tree indexing; verify PageIndex extraction quality on domain-specific PDFs (legal, medical, technical).
Wiki Markdown convention: Generated wiki is plain `.md` with cross-links; integrates with Obsidian for graph visualization but requires manual wiki maintenance workflows if hand-edited.
Concept and entity page drift: Auto-extraction and cross-document synthesis can produce over-generalized or redundant concept pages; lint and manual review of generated wiki recommended.

When to avoid it — and what to weigh

Real-Time Document Streaming — OpenKB compiles knowledge once into a static wiki. If you need live sync with evolving documents or sub-second document ingest, traditional RAG or real-time indexing is better.
Hybrid Full-Text + Vector Search — The tool deliberately avoids vector databases. If your use case requires dense similarity search or embedding-based filtering, you'll need to augment or integrate external vector systems.
Low-Latency Query Performance — Multi-turn LLM reasoning and PageIndex tree traversal add latency. Typical queries require an LLM call, not microsecond responses. Real-time autocomplete or sub-100ms QA is not the design target.
Fully Offline, Air-Gapped Deployment — OpenKB requires an external LLM API (OpenAI, Claude, Gemini, etc.) and LiteLLM communication. No built-in offline LLM support or local-only inference documented.

License & commercial use

Apache License 2.0 (Apache-2.0). Permissive OSI-approved license allowing commercial use, modification, and distribution with attribution and liability disclaimer.

Apache 2.0 permits commercial use without royalty or disclosure requirements. You may use OpenKB in proprietary products. Ensure you retain/display Apache 2.0 license terms and provide a copy. Consult legal counsel if bundling with other licenses or if your org has restrictive IP policies.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Adequate
License clarity	Clear
Deployment complexity	Low
DEV.co fit	Good
Assessment confidence	High

Security considerations

LLM API keys stored in `.env` file in project directory; ensure `.env` is in `.gitignore` and not committed. LiteLLM is pinned to a safe version (post-March 2026 security update). Input documents passed to external LLM APIs; do not add confidential/PII documents without understanding data residency. No built-in encryption for wiki at rest or in transit to LLM endpoints. Review LLM provider SLA/compliance (SOC 2, HIPAA, etc.) if handling regulated data.

Alternatives to consider

Langchain + Pinecone/Weaviate

Traditional RAG with vector database. Lower compilation overhead; real-time indexing. Trade-off: no persistent wiki, vectorless reasoning, or entity extraction. Better for high-velocity, low-latency retrieval.

Obsidian + Markdown + Manual Curation

Manual knowledge graph; no LLM compilation. Trade-off: time-intensive but full control, no API dependency, offline-capable. Best for small, curated knowledge bases.

Notion AI / Confluence AI

Closed-source cloud-hosted knowledge platforms with AI features. Trade-off: no local ownership, vendor lock-in, limited customization. Better for small teams seeking simplicity over control.

Software development agency

Build on OpenKB with DEV.co software developers

Start with `pip install openkb` and `openkb init`. Compile your first document in minutes. Free, open-source, Apache 2.0 licensed.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

MaxKB RAGHub code-graph-rag

Related on DEV.co

Explore the category and the services that help you build with it.

RAG Frameworks AI Application Development API Development Talk to DEV.co

OpenKB FAQ

Do I need a vector database?

No. OpenKB uses PageIndex tree indexing for long documents and reasoning-based retrieval. Vector embeddings are deliberately avoided to reduce infrastructure and cost complexity.

What LLM providers are supported?

Any provider supported by LiteLLM: OpenAI, Anthropic (Claude), Google Gemini, Cohere, Mistral, and 50+ others. Set via `provider/model` format in config or during `openkb init`.

Can I use this offline?

No. OpenKB requires external LLM API calls for compilation and querying. Fully offline deployment is not documented. Consider a local LLM + LiteLLM integration (requires testing).

How do I version-control the wiki?

The wiki lives in plain Markdown files in `wiki/`. Git-track the entire directory. Auto-generated pages (summaries, concepts) will have diffs on recompile; use `--dry-run` to preview changes.

Software developers & web developers for hire

From first prototype to production, DEV.co delivers software development services around tools like OpenKB. Our software development agency staffs experienced software developers and web developers for custom software development, web development, integrations, and ongoing support across rag frameworks and beyond.

Build Your Knowledge Base Today

Start with `pip install openkb` and `openkb init`. Compile your first document in minutes. Free, open-source, Apache 2.0 licensed.

Talk to DEV.co Schedule a Consultation