AutoAgent is a fully automated, zero-code framework for building LLM agents, created by the Data Science Lab at the University of Hong Kong (HKUDS) and released on GitHub, where it has collected more than 9,800 stars. Its accompanying paper describes an approach that turns plain natural-language requests into working agents.
Most agent frameworks still require developers to wire together prompts, tools and control flow by hand. AutoAgent instead lets users describe what they want in everyday language and then generates the agent for them, including tool selection and workflow design. The framework ships with ready-made agents for tasks such as deep research, file and data analysis, and web interaction, and it reported strong results on the GAIA benchmark for general AI assistants. It is an actively maintained research project, with repository updates continuing into late 2025.
AutoAgent is free and open source under the MIT License. You only pay for the LLM API calls your agents make to third-party model providers; the framework itself has no paid tier or hosted upsell.
AutoAgent suits researchers, analysts and non-specialist builders who want capable LLM agents without writing orchestration code. It is also a useful reference for agent researchers studying automated agent design, since the full generation pipeline is open to inspect and extend.