Gaia X

🚀 Gaia-x: An AI Application Framework for Enterprises

Gaia-x is an AI application framework designed for enterprises. It aims to build the next - generation AI and computer interaction interface through multi - agent, infinite canvas, and plug - in design. It supports multiple operating systems, including Windows, MacOS, and Linux, and is compatible with the Docker environment.

🚀 Quick Start

Gaia-x provides a comprehensive solution for enterprise - level AI applications. With its rich features and flexible architecture, it can be quickly deployed in various enterprise scenarios.

✨ Features

• Unified Authentication and User Management: Supports multiple authentication methods such as OAuth 2.0, LDAP, DingTalk, and Feishu.
• Unified Model and Tool Management: Provides centralized backend configuration and authorization management functions.
• Quota Management: Enables fine - grained control over user quotas and API calls.
• Enterprise Internal Application Ecosystem: Builds MCP marketplace, Agent marketplace, and Task marketplace to support applications for specific complex tasks.
• Business Reports: Generates comprehensive data analysis reports on user, model, and agent usage.

🔧 Technical Details

Project Architecture

Client

The client adopts a plug - in design and will gradually evolve towards a micro - kernel plug - in architecture similar to VS Code, supporting arbitrary extensions. The main technology stack includes:

• Framework: Electron + React
• Large Language Model UI: Ant Design X
• Text Selection Monitoring Technology: C++ (Windows), Objective - C (MacOS)

Management Backend

The management backend only exists as an interactive UI and MCP management executor and does not directly handle model calls. The main technology stack includes:

• Large Language Model Interaction: Eino framework + Self - developed multi - agent architecture
• Management UI: Ant Design Pro (Refactored based on GVA backend page code)
• API Service: Golang + Gin (Based on GVA framework)

MCP Server

• Runs in an independent sandbox environment.
• Supports implementation in Python and TypeScript.

Model Selection Reference

• Computer Operation Models: Claude Sonnet 3.5+, CogAgent, UI - TARS, OpenAI computer - use.
• Recommended ReAct Models: Claude Sonnet 3.5+, GPT 4o. (The DeepSeek v3 model performs poorly and is not recommended for critical tasks.)

📄 Acknowledgments

We would like to thank the following open - source projects and technology frameworks for their support:

• The management API is developed based on Gin - Vue - Admin, which significantly reduces our development workload for the basic management backend.
• We use the ByteDance open - source Eino framework to solve the basic integration and multi - agent architecture problems of large language model applications. We have conducted a large amount of secondary development on this basis and released it as Eino - X.
• To solve the problem of running the MCP Server on the ordinary user client, we developed support based on the project mcp - runtime.

📄 License

This project is open - sourced under the MIT License.

📚 Documentation

For more help and support regarding the Gaia - x project, please visit our official documentation or join our developer community to build the future of AI with global developers!