MaxKB: Enterprise AI Assistant with RAG, Workflows, & MCP Server Tool Use
MaxKB: Enterprise AI Assistant with RAG, Workflows, & MCP Server Tool Use
MaxKB is an open-source enterprise AI assistant, built in Python, designed to augment large language models (LLMs) with external tool interaction and robust knowledge retrieval. It supports a range of LLMs, including private models like DeepSeek and public ones such as OpenAI and Claude, and offers multi-modal input/output capabilities. With over 22,000 GitHub stars, MaxKB focuses on delivering accurate, automated AI solutions for complex business needs.
MCP Integration for Enhanced LLM Functionality
MaxKB's core strength lies in its utilization of the Model Context Protocol (MCP). This protocol is central to how MaxKB extends LLM capabilities beyond their inherent knowledge. MCP provides a standardized framework that enables LLMs to:
- Discover Tools: LLMs can dynamically identify and list available external tools and their specific functionalities.
- Understand Tool Usage: Each tool comes with a clear description of its purpose, necessary input parameters, and expected output format, allowing the LLM to correctly interpret its use.
- Execute Tools: The LLM can invoke these tools with the appropriate parameters, triggering real-world actions or data retrieval.
- Incorporate Results: Outputs from tool executions are seamlessly fed back into the LLM's context. This integration allows the LLM to make more informed decisions and generate more accurate, contextually relevant responses.
This MCP-driven approach significantly enhances MaxKB's functionality by granting LLMs access to real-world data and specialized services. It improves accuracy by reducing reliance on the LLM's internal knowledge and increases automation by orchestrating multiple tools within complex workflows.
Agentic Workflows and RAG Pipelines
MaxKB is equipped with a powerful workflow engine and a rich function library, enabling the orchestration of AI processes for complex business scenarios. This agentic workflow capability directly leverages MCP tool-use for integrating and executing external tools and services.
The platform also features a Retrieval-Augmented Generation (RAG) pipeline. This pipeline supports direct document uploading and automatic crawling of online documents, followed by automatic text splitting and vectorization for efficient knowledge indexing. The RAG pipeline significantly reduces hallucinations in LLMs, aiming for a more accurate and reliable smart Q&A experience.
Quick Setup with Docker
Getting MaxKB up and running is straightforward using Docker. This method ensures a consistent environment for deployment.
Prerequisites
- Docker installed on your system.
Installation
To install and run MaxKB, execute the following Docker command:
docker run -d --name=maxkb --restart=always -p 8080:8080 -v ~/.maxkb:/var/lib/postgresql/data -v ~/.python-packages:/opt/maxkb/app/sandbox/python-packages 1panel/maxkbThis command runs the container in detached mode, names it maxkb, configures it to restart automatically, maps port 8080, and mounts two volumes for persistent data and Python packages.