MaxKB
Open-source RAG and workflow platform for building enterprise agents
Overview
MaxKB (Max Knowledge Brain) is FIT2CLOUD's open-source platform for building RAG chatbots and agent workflows — document ingestion, MCP tool-use, and a visual workflow engine, self-hosted under GPLv3 with a paid Pro tier for SSO and higher limits.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 3 Sep 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- Up to 2 users
- 5 applications
- 50 knowledge bases
- Community support
- Unlimited users, apps, knowledge bases
- SSO (LDAP/OIDC/CAS)
- Slack/WeChat/DingTalk/Lark integration
- Voice Q&A + custom AI avatars
What you can produce with MaxKB
- Document upload + web crawling with auto-chunking and vectorization (RAG)
- Visual workflow engine with function library
- MCP tool-use support
- Model-agnostic — works with public and private LLMs
- Multi-modal input/output (text, image, audio, video)
- SSO via LDAP/OIDC/CAS (Pro edition)
- Slack, WeChat, DingTalk, Lark integrations (Pro edition)
- Self-hosted deployment (Docker), GPLv3 licensed
MaxKB is one of the more credible self-hosted RAG platforms to come out of China's open-source scene — 21,000+ GitHub stars, active releases, and a straightforward path from PDF upload to a working chatbot with citations. The workflow engine and MCP tool-use put it a notch above simple document-QA wrappers; you can chain retrieval, function calls and multi-step logic without touching code.
The catch is upkeep. GPLv3 licensing is fine for internal tools but complicates redistribution, and a security audit flagged a sandbox bypass and privilege-escalation issue in past releases — patched, but a reminder that you're running your own attack surface. The free OSS edition caps you at 2 users and 5 apps, so anything beyond a pilot pushes you toward the custom-priced Pro tier. Good fit for teams that already run Docker infrastructure and want a knowledge base off a public LLM API; not a fit for anyone wanting a managed SaaS.
What people say
MaxKB has picked up roughly 21,800 stars on GitHub as of mid-2026, built and maintained by 1Panel-dev under Hangzhou-based FIT2CLOUD — the same team behind 1Panel server management and the JumpServer bastion host. It sits in a crowded field of self-hosted RAG tools (Dify, RAGFlow, FastGPT) but distinguishes itself with a genuinely usable workflow canvas rather than just a chat-over-documents demo.
A repo analysis from The Dispatch flagged governance gaps worth knowing about before you self-host: several recent pull requests were merged without proper review approval, and the release-note process hasn't been consistently followed — not disqualifying for an open-source project, but worth factoring into how much you trust unreviewed code in a knowledge-base tool that touches your documents. GitHub's security advisories for the project also list a Python sandbox LD_PRELOAD bypass and a privilege-escalation bug, both since patched in later releases; anyone deploying MaxKB should stay current on point releases.
On the product side, the free OSS edition covers document upload, web crawling, auto-chunking/vectorization, and a function library with MCP tool support — enough to stand up a working internal assistant. It caps out at 2 users and 5 applications, though, which pushes any real team deployment toward the Pro edition (custom annual pricing, self-hosted only — there's no managed cloud option). Pro adds SSO (LDAP/OIDC/CAS), voice Q&A, custom AI avatars, and integrations with Slack, WeChat, DingTalk and Lark, plus multi-route retrieval and third-party embedding.
Documentation ships in both Chinese and English, and the model-agnostic design means you're not locked into a single LLM vendor — useful for ASEAN teams juggling regional model access and data-residency requirements. The GPLv3 license is the main friction point for companies wanting to build a commercial product on top rather than just deploy it internally.
Supported document formats cover the usual office set — Markdown, TXT, DOCX, PDF, HTML, XLSX, XLS and CSV — and the web-crawling option means you're not limited to manually uploaded files for building a knowledge base. Compared to Dify or RAGFlow, reviewers tend to describe MaxKB as the more turnkey option for non-technical admins standing up a first internal chatbot, at the cost of some flexibility that heavier frameworks offer power users.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including MaxKB's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with MaxKB unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to MaxKB directly →
Spotted something out of date? Suggest an update →
More in Agents & Automation
