Eify

A lightweight AI Agent platform — visual orchestration, multi-model support, RAG knowledge base, and MCP tool extensions. ‌The original intention behind launching this open-source initiative is to learn Vibe Coding.

中文

---

Features

Module	Description
Multi-Provider	OpenAI-compatible protocol support — DashScope, Ollama, and more
Agent Management	Visual agent creation and configuration with custom system prompts and parameters
Streaming Chat	SSE streaming responses, multi-turn conversations, context window management
Knowledge Base + RAG	Document upload, vector embedding, multi-strategy retrieval (vector/keyword/hybrid)
MCP Tools	Model Context Protocol support for extending agent capabilities
Workflow Engine	Visual drag-and-drop orchestration — conditional branches, LLM nodes, code execution
Multi-Tenant Workspace	JWT authentication + workspace-level data isolation, built for teams
Observability	Micrometer Tracing + Brave, ClickHouse + Vector log collection

Screenshots

AI Chat

Agent & Provider Management

Knowledge Base & RAG

Visual Workflow Editor

MCP Tools

Multi-Tenant Workspace

Authentication

Quick Start

Prerequisites

• Docker & Docker Compose
• JDK 21+ & Maven 3.9+ (local development only)

Docker Deploy (Recommended)

# 1. Clone
git clone https://github.com/mikewoo/eify.git
cd eify

# 2. Configure environment
cp .env.example .env
# Edit .env — set passwords and API keys

# 3. Build backend JAR
mvn clean package -DskipTests

# 4. Start all services
docker-compose -f deploy/infra/deploy/docker-compose.yml up -d

# 5. Access
# Frontend: http://localhost
# API Docs:  http://localhost:8080/doc.html

Local Development

# 1. Start dependencies (PostgreSQL + Redis)
docker-compose -f deploy/infra/deploy/docker-compose.yml up -d pgvector redis

# 2. Start backend
./start.sh dev
# Or: mvn spring-boot:run -pl eify-app -Dspring-boot.run.profiles=dev

# 3. Start frontend
cd eify-web
npm install
npm run dev

# 4. Access
# Frontend: http://localhost:5173
# API Docs:  http://localhost:8080/doc.html

Optional Components

# Logging stack (ClickHouse + Vector + Grafana)
docker-compose -f deploy/infra/deploy/docker-compose-logging.yml up -d

# Distributed tracing UI (Jaeger)
docker-compose -f deploy/optional/docker-compose-jaeger.yml up -d

# Vector database (pgvector) — already included in the main compose above

Tech Stack

Layer	Technology
Backend Framework	Spring Boot 4.0.6
ORM	MyBatis-Plus 3.5.15
Database	PostgreSQL 17 + pgvector
Cache	Redis 7
Frontend	Vue 3.5.17 + TypeScript + Vite
UI Framework	Element Plus 2.10.2
State Management	Pinia 2.3.1
Workflow UI	Vue Flow
Tracing	Micrometer Tracing + Brave
Log Storage	ClickHouse 25 + Vector 0.54
Containerization	Docker + Docker Compose

Project Structure

eify/
├── eify-app/              # Application entry point
├── eify-auth/             # Auth & workspace (JWT, user management, multi-tenant)
├── eify-provider/         # LLM provider adapters
├── eify-agent/            # Agent creation & configuration
├── eify-chat/             # Chat engine (SSE streaming)
├── eify-knowledge/        # Knowledge base + RAG retrieval
├── eify-mcp/              # MCP tool protocol
├── eify-workflow/         # Workflow engine
├── eify-common/           # Shared utilities
├── eify-web/              # Vue 3 frontend
├── docs/                  # Project documentation (Chinese)
├── deploy/                # Deployment configs (Docker, K8s, SQL)
├── scripts/               # Utility scripts
├── start.sh               # Start script
├── stop.sh                # Stop script
└── .env.example           # Environment variable template

Architecture

┌─────────────┐     ┌─────────────┐     ┌──────────────┐
│   Vue 3     │────▶│  Spring Boot│────▶│ PostgreSQL 17│
│   Frontend  │     │  REST API   │     │  + pgvector  │
└─────────────┘     └──────┬──────┘     └──────────────┘
                           │
               ┌───────────┼───────────┐
               │           │           │
               ▼           ▼           ▼
        ┌──────────┐ ┌──────────┐ ┌──────────┐
        │  Redis   │ │ LLM/    │ │ ClickHouse │
        │  Cache   │ │ Vector   │ │ LLM APIs │
        └──────────┘ └──────────┘ └──────────┘

• Auth: Stateless JWT + ThreadLocal context propagation
• Multi-tenant: All business data isolated by workspace_id, enforced via WorkspaceGuard
• Streaming: SSE long connections with per-token LLM push
• Observability: Auto-generated TraceId/SpanId via Micrometer Tracing + Brave, ClickHouse-backed logs (structured JSON)

Documentation

Note: Detailed docs are currently in Chinese. English translations are planned.

Document	Description
ARCHITECTURE.md	Architecture & coding conventions
API-SPEC.md	API design specification
AUTH-WORKSPACE.md	Auth & multi-tenant workspace
DATABASE.md	Database design (PostgreSQL + ClickHouse)
LOGGING.md	Logging system guide
WORKFLOW.md	Workflow engine design
DEPLOYMENT.md	Deployment & CI/CD

Roadmap

• Ollama local model adapter
• Anthropic Claude adapter
• Chat history export
• Agent template marketplace
• Frontend internationalization (i18n)
• Frontend test coverage
• English documentation

Contributing

Issues and pull requests are welcome. Before contributing, please read:

• Contributing Guide (Chinese)
• Code of Conduct (Chinese)

License

MIT License © 2026 mikewoo

---

_{Made by mikewoo | MIT License}