Empower every employee to converse with business systems through natural language

---

🎯 What is EvolveHub?

🧬	EvolveHub = Enterprise AI Middleware Platform

EvolveHub is an enterprise AI middleware service system built on AgentScope-Java. Through a unified AI capability access layer, departments can quickly integrate intelligent conversation, knowledge base retrieval and other AI capabilities by connecting MCP services or A2A protocol + Skills packages, without redundant development.

🪄 Core Values

Unified Access	Knowledge Accumulation	Intelligent Conversation	Security & Control
Fast integration via MCP/A2A	Centralized knowledge base, intelligent retrieval	Natural language interaction, tool execution, task automation	Multi-level permissions, on-premise deployment

---

🏗️ Platform Architecture

graph TB
    subgraph "👤 User Layer"
        WEB[Web Console]
        API[OpenAPI]
    end

    subgraph "🌐 Access Layer"
        GW[API Gateway<br/>Load Balance / Routing / Rate Limit / Auth]
    end

    subgraph "⚙️ Service Layer"
        AUTH["Auth Service :8081<br/>Authentication (Sa-Token)"]
        AI["AI Platform :8082<br/>AI Middleware Core"]
        ADMIN["Admin Service :8083<br/>Admin Management"]
    end

    subgraph "🤖 Agent Layer"
        REACT[ReActAgent<br/>Reasoning Engine]
        TOOL[Toolkit<br/>MCP Tools]
        MEM[Memory<br/>Memory Management]
        GUARD[ToolGuard<br/>Security Guardrail]
    end

    subgraph "🔌 Extension Services"
        CRON[Cron Scheduler]
        APRV[Approval Workflow]
        SKILL[Skill Manager]
        STATS[Token Stats]
    end

    subgraph "💾 Data Layer"
        PG[("MySQL<br/>Business Data")]
        RD[("Redis<br/>Cache / Short-term Memory")]
        MV[("Milvus<br/>Vector Store")]
        MI[("MinIO<br/>File Storage")]
    end

    WEB --> GW
    API --> GW
    GW --> AUTH
    GW --> AI
    GW --> ADMIN
    AI --> REACT
    REACT --> TOOL
    REACT --> MEM
    REACT --> GUARD
    AI --> CRON
    AI --> APRV
    AI --> SKILL
    AI --> STATS
    AUTH --> PG
    AUTH --> RD
    AI --> PG
    AI --> RD
    AI --> MV
    ADMIN --> MI

    style GW fill:#FF6B6B,color:#fff
    style AI fill:#4ECDC4,color:#fff
    style REACT fill:#9B59B6,color:#fff
    style TOOL fill:#6DB33F,color:#fff
    style GUARD fill:#E74C3C,color:#fff

Loading

---

✨ Phase 1 Core Features

💬 Smart Conversation Multi-turn dialogue + context ReAct reasoning engine SSE streaming output Auto intent routing	📚 Knowledge Base 4 tiers: Global / Dept / Project / Sensitive RAG retrieval-augmented generation Permission filtering + semantic search Auto document chunking & vectorization	🧠 Memory System Short-term (Redis, 30min TTL) Long-term (Mem0 + Milvus) User config (MinIO) Sleep consolidation
🔧 Tool Execution MCP protocol integration Built-in CLI toolset Security guardrail (ToolGuard) Injection / traversal detection	🤖 Model Management Cloud: Qwen / OpenAI / Gemini / Claude Local: Ollama / vLLM Embedding: Qwen3 / bge Web UI dynamic switching	⏰ Cron Scheduler Cron expression scheduling Heartbeat + zombie task recovery Concurrency control Full status tracking
✅ Approval Workflow High-risk operation interception Web confirm + admin approval Auto expiry (24h default) GC mechanism	⚡ Skills Extension Built-in: DOCX / PDF / PPTX / XLSX Upload + security scan Lifecycle management Sandbox isolation	📊 Token Statistics By user / dept / model stats Budget control + alerts Usage report export Cost trend analysis

---

🔐 Role-Based Access Control

graph TB
    L1["👑 Super Admin<br/>Full Platform · System Config · Budget"]
    L2["🏢 Executive<br/>Cross-Dept View · Strategic Decisions"]
    L3["📂 Dept Manager<br/>Dept Management · Data Insights"]
    L4["🔧 Admin<br/>Project Mgmt · Agent Collaboration"]
    L5["👤 Employee<br/>Personal Chat · Knowledge Search"]

    L1 --> L2 --> L3 --> L4 --> L5

    style L1 fill:#FF6B6B,color:#fff
    style L2 fill:#4ECDC4,color:#fff
    style L3 fill:#45B7D1,color:#fff
    style L4 fill:#9B59B6,color:#fff
    style L5 fill:#6DB33F,color:#fff

Loading

---

🚀 Use Cases

Scenario	Description	Key Capability
💬 Smart Customer Service	AI understands business, auto-queries orders, handles tickets	Chat + MCP Tools
📊 Data Assistant	Natural language database queries, report generation	Knowledge Base + RAG
🔧 Ops Assistant	AI executes operations, auto-troubleshoots	Tool Execution + Approval
📋 Workflow Approval	Intelligent approval understanding, decision support	Workflow + Guardrails
🎓 Training Tutor	Q&A based on enterprise knowledge base	Knowledge Base + Memory
📈 Scheduled Reports	Auto-generate daily/weekly reports, push notifications	Cron + Skills

---

🔌 Integration Methods

Method 1: MCP Protocol

Configure your MCP service endpoint, platform auto-discovers and loads tools:

# evolvehub-config.yaml
mcp:
  servers:
    - name: "company-erp"
      endpoint: "https://erp.company.com/mcp"
      auth:
        type: "bearer"
        token: "${ERP_API_TOKEN}"

Method 2: A2A Protocol

Register your Agent to A2A network for multi-agent collaboration:

a2a:
  registry: "nacos://localhost:8848"
  agents:
    - name: "order-agent"
      capability: "Order Query & Processing"
    - name: "inventory-agent"
      capability: "Inventory Management"

Method 3: Skills Packages

Import pre-built enterprise skill packages for instant business capabilities:

skills:
  - name: "database-query"
    version: "1.0.0"
  - name: "report-generator"
    version: "2.1.0"

---

🛠️ Tech Stack

Component	Technology	Version	Description
Agent Framework	AgentScope-Java	1.0.11	Core Agent Engine (ReAct)
Backend Framework	Spring Boot	3.2+	Microservice Framework
JDK	OpenJDK	21	LTS
Authentication	Sa-Token	1.37+	Lightweight Auth Framework
Cache	Redis	8.x	Short-term Memory + Cache + Queue
Vector Database	Milvus	2.x	Knowledge Base Vector Storage
Relational Database	MySQL	8.0+	Business Data Storage
File Storage	MinIO	Latest	Document / User Config Storage

---

🆚 Comparison with Traditional Solutions

Dimension	Traditional AI Development	EvolveHub
Development Cost	🔴 High (needs AI engineers)	🟢 Zero-code config
Deployment Time	🔴 Weeks/Months	🟢 Minutes
Business Adaptation	🔴 Custom development	🟢 MCP/A2A plug-and-play
Knowledge Building	🔴 Static Prompts	🟢 Auto-evolution + sleep consolidation
Security Control	🔴 Manual review	🟢 Guardrails + approval workflow
Maintenance Cost	🔴 Continuous investment	🟢 Self-adaptive optimization

---

📦 Deployment Options

Deployment Mode	Use Case	Features
🐳 Docker	Quick trial, test environments	One-click startup
☸️ Kubernetes	Production, high availability	Elastic scaling
🏢 On-Premise	Data-sensitive, compliance	Full control

Docker Quick Start

# Pull image
docker pull evolvehub/server:latest

# Start service
docker run -d \
  --name evolvehub \
  -p 8080:8080 \
  -v ./config:/app/config \
  evolvehub/server:latest

# Visit http://localhost:8080 to start using

---

📈 Roadmap

%%{init: {'theme': 'base', 'themeVariables': { 'primaryColor': '#4ECDC4'}}}%%
timeline
    title EvolveHub Roadmap
    section Phase 1 · Core Capabilities
        M1 Infrastructure : Microservices · Database · Middleware
        M2 Core Features : Chat System · Knowledge Base · Memory
        M3 Tool Integration : MCP Tools · Security Guardrails · Approval
        M4 Extension System : Cron Tasks · Skills · Security Scan
        M5 Operations : Token Stats · Workspace · Hot Reload
        M6 Admin Console : User Mgmt · KB Mgmt · Monitoring
        M7 Launch : Testing · Performance · Canary Release
    section Phase 2 · Extended Capabilities
        M8 Agent Collab : Multi-Agent · Task Decomposition · Zero-Downtime
        M9 A2A Protocol : Cross-Dept Communication · Access Control
        M10 Online Training : Feedback Optimization · Evaluation
        M11 Data Collection : Auto Collection · Knowledge Graph

Loading

---

🤝 Join the Community

📱 Scan to Join DingTalk Group

Product Inquiry · Technical Discussion · Feedback

---

📄 License

---

Made with ❤️ by the EvolveHub Team