## Introduction The code generation task is a complex one that requires good understanding of the language and syntax. However, LLMs have shown to be effective in generating code from English prompts, but this progress has not extended to low-resource languages like Bengali. The research team developed a new framework for generating code in Bengali called BanglaCodeAct. This framework uses multilingual agents and iterative self-correction to improve the accuracy of code generation. The project was presented at the BLP-2025 conference, where researchers presented their results and discussed the potential applications of the framework. ## Results The results of the project show that the BanglaCodeAct framework can generate Bengali code with high accuracy. The tests were conducted on the mHumanEval platform, where researchers evaluated the performance of the framework on a dataset of Bengali code. The results show that the BanglaCodeAct framework can achieve an accuracy of 94% on the development set and 71.6% on the blind test set. These results establish a new benchmark for Bengali-to-Python translation and highlight the potential of agent-based reasoning for reliable code generation in low-resource languages. ## Conclusion The BanglaCodeAct project represents an important milestone in code generation for low-resource languages. The use of multilingual agents and iterative self-correction allows the framework to improve the accuracy of code generation. Researchers hope that the project can be used to improve understanding of language and syntax in low-resource languages, opening up new possibilities for application of the framework in various fields.