English | 简体中文 | 繁體中文 | 日本語 | Русский
TGO is an open-source AI agent customer service platform dedicated to helping enterprises "Build AI Agent Teams for Customer Service". It integrates multi-channel access, agent orchestration, knowledge base management (RAG), and human agent collaboration.
Run the following command on your server to check requirements, clone the repository, and start the services:
REF=latest curl -fsSL https://raw.githubusercontent.com/tgoai/tgo/main/bootstrap.sh | bash
For users in China (using Gitee and Aliyun mirrors):
REF=latest curl -fsSL https://gitee.com/tgoai/tgo/raw/main/bootstrap_cn.sh | bash
For more details, please visit the Documentation.
- Multi-Agent Support - Configure multiple AI agents for different business scenarios
- Multi-Model Integration - Connect with various LLM providers (OpenAI, Anthropic, etc.)
- Streaming Response - Real-time AI responses via SSE for smooth conversation experience
- Context Memory - Maintain conversation history for coherent dialogue
- Document Knowledge Base - Upload documents to enhance AI response accuracy
- Q&A Knowledge Base - Create question-answer pairs for quick knowledge expansion
- Website Knowledge Base - Crawl website content to keep information up-to-date
- Smart Retrieval - Vector-based semantic search for precise answers
- Tool Store - Rich library of MCP tools, enable on demand
- Custom Tools - Project-level tool configuration and management
- OpenAPI Schema - Auto-parse schemas to generate interactive forms
- Web Widget - Embeddable chat widget for websites
- WeChat Integration - Official Account and Mini Program support
- Unified Management - Manage all channels from a single dashboard
- WuKongIM Integration - Stable and reliable instant messaging
- WebSocket Connection - Efficient bidirectional communication
- Message Sync - Read/unread status, delivery confirmation
- Rich Media - Support for text, images, files and more
- Smart Handoff - Seamlessly transfer to human agents when needed
- Visitor Management - Collect visitor info, assign sessions, track history
- Agent Workspace - Unified interface for human agents
- Structured Display - Render orders, products, logistics as beautiful cards
- Rich Components - Order cards, logistics tracking, product display, price comparison
- Action Protocol - Standardized URI protocol for interactions
| Repository | Description | Tech Stack |
|---|---|---|
| tgo-ai | AI/ML operations service for managing agents, tool bindings, knowledge bases, and usage analytics | Python / FastAPI |
| tgo-api | Core business logic service handling user management, visitor tracking, assignment, and communication | Python / FastAPI |
| tgo-cli | CLI tool & MCP Server enabling AI agents to execute customer service operations with 40+ built-in tools | TypeScript / Node.js |
| tgo-device-agent | Embedded agent running on managed devices providing file and shell capabilities via TCP JSON-RPC | Go |
| tgo-device-control | Device control service managing TCP/JSON-RPC connections for remote device management with MCP Agent | Python / FastAPI |
| tgo-platform | Multi-channel message intake service supporting WeChat, Feishu, DingTalk, Telegram, Slack, email, etc. | Python / FastAPI |
| tgo-plugin-runtime | Plugin lifecycle management and execution service with dynamic tool synchronization | Python / FastAPI |
| tgo-rag | RAG service providing document processing, hybrid semantic/full-text search, and async processing | Python / FastAPI |
| tgo-web | Admin frontend with real-time chat, AI agent management, knowledge base, and MCP tool integration | TypeScript / React 19 |
| tgo-workflow | AI Agent workflow execution engine supporting DAG topology with LLM, API, condition, and tool nodes | Python / FastAPI |
| Repository | Description | Tech Stack |
|---|---|---|
| tgo-widget-js | Embeddable customer service chat widget (Intercom-style) for websites | TypeScript / React 18 |
| tgo-widget-ios | Native iOS customer service chat SDK with SwiftUI views and UIKit bridging | Swift / SwiftUI |
| tgo-widget-flutter | Cross-platform customer service chat widget for iOS and Android | Dart / Flutter |
| tgo-widget-cli | Visitor-facing CLI tool & MCP Server providing customer service interface | TypeScript / Node.js |
| tgo-widget-miniprogram | WeChat Mini Program chat widget component with AI streaming responses and Markdown rendering | TypeScript |
Dashboard  |
Agent Orchestration  |
Knowledge Base  |
Q&A Debugging  |
MCP Tools  |
Platform Admin  |
- CPU: >= 4 Core
- RAM: >= 8 GiB
- OS: macOS / Linux / WSL2