In this paper, we introduce
FullStack-Agent, a unified system that combines a multi-agent full-stack development framework equipped with efficient coding and debugging tools (
FullStack-Dev), an iterative self-improvement method that improves the abilities of LLMs through repository augmentation and back-translation (
FullStack-Learn), and a full-stack development benchmark that comprehensively evaluates frontend, backend, and database functionalities (
FullStack-Bench).
Instructions for installation and running of the three components are in the following documents:
Experimental results of FullStack-Dev on
FullStack-Bench compared to popular baseline methods are shown below:
Using more templates result in better performance in most of the metrics, which might be due to the fact that with more templates to choose from, the agent can find the most appropriate and easy-to-work-with templates, thus making the development process smoother.
Experimental results of FullStack-Learn tested on with FullStack-Dev on FullStack-Bench are as follows:
1@misc{lu2026fullstackagentenhancingagenticfullstack,
2 title={FullStack-Agent: Enhancing Agentic Full-Stack Web Coding via Development-Oriented Testing and Repository Back-Translation},
3 author={Zimu Lu and Houxing Ren and Yunqiao Yang and Ke Wang and Zhuofan Zong and Mingjie Zhan and Hongsheng Li},
4 year={2026},
5 eprint={2602.03798},
6 archivePrefix={arXiv},
7 primaryClass={cs.SE},
8 url={https://arxiv.org/abs/2602.03798},
9}
