AutoGen Studio: A No-Code Developer Tool for Building and Debugging Multi-Agent Systems

多智能体系统，即多个智能体（生成型AI模型+工具）协同合作，正在成为解决许多领域中长期、复杂任务的有效模式。然而，对于大多数开发人员来说，指定它们的参数（例如模型、工具和编排机制等）以及调试它们仍然具有挑战性。为了解决这个问题，我们提出了AUTOGEN STUDIO，这是一个无代码开发工具，用于基于AUTOGEN框架快速原型设计、调试和评估多智能体工作流。AUTOGEN STUDIO提供了一个Web界面和一个Python API，用于使用声明性（基于JSON的）规范表示LLM启用的智能体。它提供了一个直观的拖放UI，用于智能体工作流规范，交互式评估和调试工作流，以及可重用智能体组件的库。我们强调了无代码多智能体开发工具的四个设计原则，并在https://github.com/microsoft/autogen/tree/main/samples/apps/autogen-studio上贡献了一个开源实现。

AUTOGEN STUDIO: A No-Code Developer Tool for Rapid Prototyping and Debugging of Multi-Agent Workflows

AUTOGEN STUDIO provides a no-code solution for rapidly prototyping, debugging, and evaluating multi-agent workflows using a declarative (JSON-based) specification and an intuitive drag-and-drop UI.

The tool offers a web interface and a Python API for representing LLM-enabled agents, interactive evaluation and debugging of workflows, and a gallery of reusable agent components. It follows four design principles for no-code multi-agent developer tools. The open-source implementation is available on GitHub.

Related work includes research on multi-agent systems, declarative programming, and no-code development tools. Some relevant papers include 'Multi-Agent Systems: A Survey' by Weiss, 'A Declarative Language for Multi-Agent Systems' by Pokahr et al., and 'No-Code Development Platforms: A Systematic Mapping Study' by Mendes et al.