A general AI agent framework designed to systematically build a complete agent capability closed loop, from perception to collaboration, for complex enterprise task scenarios. It targets developers and organizations seeking to accelerate intelligent upgrades and industry collaboration by providing a robust, extensible, and secure agent engine.

How It Works

JoinAI-Agent employs a frontend-backend separation architecture. The backend utilizes a DAG execution engine, integrating multiple specialized sub-agents (e.g., information integration, code expert, report generation) and a rich tool ecosystem (search, file processing). This modular design facilitates integration and maintenance, while a secure sandbox environment (E2B) ensures safe code execution for tasks like Shell commands, file operations, and browser automation.

Quick Start & Requirements

• Primary Install: Deploy using Docker Compose (docker-compose up -d).
• Prerequisites: Docker, Docker Compose, Python 3.12+ (local development), Node.js 20+ (local development).
• Configuration: Requires setting environment variables in .env files for backend and frontend services. This includes API keys for LLMs (e.g., OpenAI, SiliconFlow), search tools (JINA, SERPER, Bocha), and the E2B sandbox. Specific configurations are needed for A2A agents (e.g., image parsing with VL models).
• Links: E2B API: https://e2b.dev/, JINA: https://jina.ai/, SERPER: https://serper.dev, Bocha: https://bocha.cn/.

Highlighted Details

• Achieved a top score of 90.70 on the GAIA benchmark.
• Offers an end-to-end product with one-click deployment, supporting secondary development.
• Integrates standardized MCP and A2A protocols for pluggable tool and agent extensions.
• Features a secure E2B sandbox environment for code execution and browser automation capabilities.
• Includes pre-built sub-agents for reporting, research, and coding.

Maintenance & Community

The project welcomes contributions. Specific details on core maintainers, community channels (like Discord/Slack), or a public roadmap are not detailed in the README.

Licensing & Compatibility

Licensed under the Apache License 2.0. This license is permissive and generally compatible with commercial use and linking in closed-source projects.

Limitations & Caveats

This project is under active development, meaning APIs and configurations are subject to change. Users should consult the latest documentation and code updates. Planned features include enhanced reasoning modes, expanded tool support, and multi-agent memory.