Loading repository data…
Loading repository data…
taishan666 / repository
MaxKB4j is an open-source LLMOps platform for LLM workflow applications and RAG developed based on the Java language. The project mainly draws on MaxKB, Dify and FastGPT, and combines the advantages of the two into one project. It is redesigned and developed using the high-performance, high-stability and secure reliable JAVA language.
A transparent discovery signal based on current public GitHub metadata.
This score does not audit code, security, maintainers, documentation quality, or suitability. Verify the repository and its current documentation before adoption.
MaxKB4j = Max Knowledge Brain for Java A ready-to-use, secure, model-agnostic RAG (Retrieval-Augmented Generation) + LLM workflow engine, purpose-built for enterprise-grade intelligent Q&A systems. Widely used in scenarios such as intelligent customer service, internal enterprise knowledge bases, data analysis, academic research, and education.
In today's AI application boom, are you facing these challenges?
MaxKB4j Provides You with a One-Stop Solution:
Built on Java 21 + Spring Boot 3 + Virtual Threads, perfectly integrating RAG (Retrieval-Augmented Generation) with visual workflow. Empower your applications with AI capabilities of "understanding, reasoning, and execution" without modifying your existing systems.
| Feature Category | Detailed Description |
|---|---|
| ⏰ Triggers | • Scheduled Task Trigger: Supports configuring Cron expressions or visual timeline for unattended automation of agents and tools (e.g., daily automatic data report generation, scheduled competitor information crawling).• Event Callback Trigger: Supports Webhook integration with external system events for real-time response (e.g., automatically trigger customer profiling Agent when new leads are added in CRM, trigger alert notifications when database data changes). |
| 🔍 Out-of-the-Box Knowledge Base Q&A | • Supports uploading local documents (PDF/Word/TXT/Markdown, etc.)• Supports automatic web content crawling• Supports custom workflow knowledge base writing• Automatically handles: text chunking → vectorization → storage in vector database → RAG pipeline construction• Significantly reduces LLM "hallucinations", improving answer accuracy and reliability |
| ⚡ High Concurrency & High Performance |
| • Built on Java 21 + Spring Boot 3 + Virtual Threads (Project Loom), fully leveraging modern JVM's lightweight concurrency capabilities for significantly improved throughput and response speed.• Adopts reactive programming model (Reactor) and asynchronous non-blocking I/O, effectively handling thousands of concurrent requests with lower resource usage and lower latency.• Built-in multi-level caching mechanism to accelerate knowledge retrieval and model invocation chains. |
| 🌐 Model-Agnostic & Flexible Integration | Supports various mainstream large language models, including:• Local Private Models: DeepSeek-R1, Llama 3, Qwen 2, etc. (via Ollama / Xorbits Inference / LocalAI)• Chinese Public Models: Tongyi Qianwen, Tencent HunYuan, ByteDance Doubao, Baidu Qianfan, Zhipu GLM, Kimi, DeepSeek, etc.• International Public Models: OpenAI (GPT), Anthropic (Claude), Google (Gemini) |
| ⚙️ Visual Workflow Orchestration | • Built-in low-code AI workflow engine, supports conditional branching, function calling, multi-turn conversation memory, and long-term memory• Provides a rich built-in function library (HTTP requests, database queries, time processing, regex extraction, etc.)• Suitable for complex business scenarios: customer support ticket generation, data report interpretation, internal policy Q&A, etc. |
| 🤝 Multi-Agent Collaboration | • Built-in Multi-Agent collaboration framework, supports multiple specialized AI Agents working in parallel or sequentially• Each Agent can be configured with independent roles (e.g., data analyst, code reviewer, customer service specialist), dedicated knowledge bases and toolsets• Supports dynamic task distribution and context-aware Agent routing; complex tasks are automatically decomposed, assigned, and aggregated (e.g., user question → requirement understanding Agent → data query Agent → report generation Agent)• Provides inter-Agent communication mechanism and shared memory bus, ensuring information consistency and collaboration coherence• Suitable for advanced scenarios: cross-department process automation, end-to-end product design, joint fault diagnosis, etc. |
| 🧩 Seamless Integration into Existing Systems | • Provides RESTful API and frontend embedding components (iframe / Web SDK)• No need to modify existing systems, integrate intelligent Q&A capabilities in 5 minutes• Provides OpenAI-compatible dialogue interface• Provides agent integration via stream_http MCP-compatible interface |
| 🤖 Skill Tools | • Supports the MCP protocol, enabling AI to understand code context, project structure, and dependencies• Supports local code function programming tool calls• Supports HTTP interface tool calls• Supports Claude SKILLS skill invocation |
| 🎙️ Multimodal Extensions | • Speech Recognition (ASR), Speech Synthesis (TTS)• Image Recognition (OCR), Image Generation (Stable Diffusion) |
| 🔒 User Permission Management | • Fine-grained permission control (application / knowledge base / tool / model)• Audit logs, authentication and authorization (based on Sa-Token) |
| 🌱 Ecosystem Extensions (Extensibility & Out-of-the-Box) | • Rich Agent template library: Provides dozens of pre-built Agent templates (e.g., customer service assistant, data analyst, code mentor, meeting note taker), one-click enable, quick adaptation to business scenarios.• Flexible plugin tool marketplace: Supports dynamic loading of functional modules through plugin mechanism, including:✅ Data connectors (MySQL, PostgreSQL, MongoDB, etc.)✅ Third-party service integrations (Feishu, DingTalk, WeCom)✅ Web search tools (Google Search, SearchApi, SearXNg, etc.) |
# Start the application
java -jar maxkb4j-start.jar
docker run --name maxkb4j -d --restart always -p 8080:8080 -e SPRING_DATASOURCE_URL=jdbc:postgresql://localhost:5432/MaxKB4j -e SPRING_DATASOURCE_USERNAME=postgres -e SPRING_DATASOURCE_PASSWORD=123456 -e SPRING_DATA_MONGODB_URI=mongodb://admin:123456@localhost:27017/MaxKB4j?authSource=admin registry.cn-hangzhou.aliyuncs.com/tarzanx/maxkb4j
-p 8080:8080 is the host port, the second 8080 is the container port-e SPRING_DATASOURCE_URL=jdbc:postgresql://localhost:5432/MaxKB4j -e SPRING_DATASOURCE_USERNAME=postgres -e SPRING_DATASOURCE_PASSWORD=123456 are PostgreSQL database connection parameters, can be modified as needed-e SPRING_DATA_MONGODB_URI=mongodb://admin:123456@localhost:27017/MaxKB4j?authSource=admin is the MongoDB connection parameter, can be modified as needed# See docker-compose.yml example in project root directory
docker-compose up -d
Sealos servers are located overseas, no need to handle network issues separately, supports high concurrency & dynamic scaling.
Click the button below for one-click deployment:
admintarzan@123456On first launch, the database (PostgreSQL + MongoDB) will be automatically initialized. Please ensure the ports are not occupied.
| Category | Technology |
|---|---|
| Backend | Java 21, Spring Boot 3, Sa-Token (Authentication) |
| AI Framework | LangChain4j |
| Vector Database | PostgreSQL 15 + pgvector |
| Full-Text Search | MongoDB 5.0+ |
| Caching | Caffeine |
| Frontend | Vue 3, Node.js v20.16.0 |
| Script Sandbox | groovy-sandbox |
demodemo@123456We welcome community contributions! If you have suggestions, bug reports, or new feature requests, please submit them via Issue or directly submit a Pull Request.
| Category | Description |
|---|---|
| 🎯 How to Contribute | Fix bugs, develop new features, improve documentation, write tests, or optimize UI/UX. |
| 📋 Process | Fork project → Create branch → Commit changes → Push to DEV branch → Open Pull Request. |
| 🎨 Standards | Follow Alibaba Java Coding Guidelines, include unit tests, and update documentation. |
🌟 Open source is not easy, persistence is harder MaxKB4j is maintained by individual developers and community members. Your support will be directly used for server costs, token testing consumption, API testing, bug fixes, and new feature development!
| Tier | Amount | Core Benefits | Target Audience |
|---|---|---|---|
| ☕ Coffee Support | ¥10 | • Add the author on WeChat: vxhqqh• Join the core discussion group (mention "sponsored")• Priority notification of project updates | Individual developers who recognize project value |
| 📚 Learning Member | ¥99 | • All Coffee Support benefits • Free access to 《👉 Knowledge Planet🔥》• Priority answers to questions in the Planet | Developers who want to learn in depth |
| 🏢 Enterprise Partner | ¥799 | • All Advanced Developer benefits • One-time access to frontend source code • Issue after-sales technical support (price continues to ris |