MaxKB

MaxKB, short for Max Knowledge Brain, is an open-source platform for building enterprise-grade agents. It gathers retrieval, workflow orchestration and tool use into one product, so those pieces do not have to be assembled and maintained separately.

It is applied in intelligent customer service, corporate internal knowledge bases, academic research and education. Grounding answers in material you supply is what reduces hallucination from the underlying model and makes the question and answer experience usable.

Five capabilities describe what the platform covers.

• RAG pipeline: documents are uploaded directly or crawled from online sources, then split and vectorized automatically.
• Agentic workflow: a workflow engine, a function library and MCP tool use orchestrate processes for complex business scenarios.
• Integration: zero-coding integration adds question answering to third-party business systems that already exist.
• Model-agnostic: private models such as DeepSeek, Llama and Qwen sit alongside public ones like OpenAI, Claude, Gemini and MiniMax.
• Multi modal: text, image, audio and video are natively supported as both input and output.

A single Docker command starts the container and serves the web interface on port 8080 with a default administrator account. Underneath it runs Vue.js on the front end, Python with Django on the back, LangChain for the model layer and PostgreSQL with pgvector for storage. An offline installation guide covers environments that cannot pull the image, and the whole thing is licensed under GPLv3.