Generated by Qwen Image.

We introduce Science-Star, the open-source framework built to revolutionize how we build, extend, and experiment scientific AI agents. Frustrated by the limitations of existing tools, we engineered Science-Star to be intuitive yet powerful. At its core, Science-Star combines a robust ReAct-based engine with essential features like Planning, Action, Memory and Reflection, integrated HLE dataset support, and powerful visualization tools. It's an all-in-one platform designed to get your ideas from concept to reality, faster than ever. Whether you are a seasoned researcher or a curious developer, Science-Star provides the tools you need to push the boundaries of science. The next breakthrough is waiting. Let's build it together. Join the Science-Star community today!

Also check out Awesome-Agent-Craft: Our curated collection of papers and benchmarks on unlocking the potential of Scientific AI Agents.

• Integrated Visualization: An end-to-end, extensible visualization suite powered by streamlit. It streamlines the entire workflow from data inspection and real-time experiment monitoring to results logging and analysis.

• Plug-and-Play Modularity: Core components (dataloader, memory, planner, tool, evaluator) are designed with well-defined interfaces. This modularity enables effortless substitution and customization.

• Scientific Extensibility: Science-Star has built-in support for advanced retrieval and literature-based Retrieval-Augmented Generation (RAG).

• More Scientific Extensibility: We will add more tools for seamless integration of specific scientific tools (e.g., Chemistry, Biology).

• More Architectures Support: We will support more agent architectures outside the ReAct framework.

• More Tasks and Datasets Support: We will support more agent datasets and tasks, like GAIA etc.

• Brief Project Structure
• Environment Setup
• Quick Start: Try o4-mini + ReAct on HLE-Small

Additional resources are available in the codebase:

• Example tools: science_star/tools/
• Data preprocessing: science_star/data_utils
• Visualization: visualization/vis_xx

Using o4-mini-2025-04-16 as our base model, we have achieved state-of-the-art (SOTA) results on a small-scale HLE dataset by implementing an end-to-end pipeline that leverages the ReAct framework with integrated planning, action, memory and reflection modules. The project requires further testing and refinement. We invite the open-source community to join us in shaping the future of this work. Let's build together!

We welcome all forms of feedback! Please raise an issue for bugs, questions, or suggestions. This helps our team address common problems efficiently and builds a more productive community.

Join our community: Connect with other users and our development team in our [WeChat group](https://github.com/Melmaphother/Science-Star/blob/main/assets/wechat.j p g).

Student Contributors: Daoyu Wang, Qingchuan Li, Tian Gao, Shuo Yu, Xiaoyu Tao, Ze Guo

Supervisors: Qi Liu, Mingyue Cheng

Affiliation: State Key Laboratory of Cognitive Intelligence, University of Science and Technology of China

We extend our gratitude to OAgent for providing the OAgent and hard work in engineering. We are also thankful to the smolagent team for their fundamental support. Lastly, we deeply appreciate the insightful discussions and contributions from Daoyu Wang, Qingchuan Li, Tian Gao, Shuo Yu, Xiaoyu Tao, Ze Guo.

Science-Star

@misc{Science-Star,
author = {Daoyu Wang, Qingchuan Li, Tian Gao, Shuo Yu, Xiaoyu Tao, Mingyue Cheng, Qi Liu},
title = {Science-Star: An Open Platform for Building, Extending, and Experimenting with Scientific Agents.},
year = {2025},
organization = {GitHub},
url = {https://github.com/Melmaphother/Science-Star},
}