Yuma: Generation of Multi-Agent Workflows from Natural Language

• André Neves UFCG
• Daniel Mantovani UFCG
• Caíque Campelo UFCG
• Gabriel Souza UFCG
• Débora Souza UFCG
• Beatriz Furtado UFCG
• Leandro Balby UFCG
• Eliane Araújo UFCG

DOI: https://doi.org/10.5753/webmedia_estendido.2025.16421

Abstract

Large Language Models (LLMs) have enabled the translation of natural language into functional systems. However, creating multiagent systems based on these models still requires advanced technical knowledge. In this work, we present Yuma, a multi-agent tool that simulates software engineering roles — Requirements Engineer and Software Architect — to collaboratively generate multi-agent systems. This approach translates users’ informal requirements into deployable workflows, with human supervision at each step, ensuring transparency and validation. The tool coordinates requirements elicitation, architecture definition, and system generation, reducing the technical barrier to creating solutions. Finally, we present the tool’s architecture, the interaction flow, and a usage example, demonstrating how this approach democratizes access to multi-agent technologies, making them accessible and transparent to non-specialized users.

Keywords: Sistemas Multiagentes, Geração Automática de Sistemas Multiagentes, Orquestração de Agentes, Vibe Coding

References

X. Du et al., “Evaluating large language models in class-level code generation,” in Proc. IEEE/ACM 46th Int. Conf. on Software Engineering, New York, NY, USA, ACM, 2024, pp. 1–13.

J. Jiang et al., “A survey on large language models for code generation,” arXiv preprint arXiv:2406.00515, 2024.

D. Souza et al., “Code Generation with Small Language Models: A Deep Evaluation on Codeforces,” arXiv preprint arXiv:2504.07343, 2025.

C. E. Jimenez et al., “Swe-bench: Can language models resolve real-world github issues?,” arXiv preprint arXiv:2310.06770, 2023.

L. Belzner, T. Gabor, and M. Wirsing, “Large language model assisted software engineering: prospects, challenges, and a case study,” in Int. Conf. on Bridging the Gap between AI and Reality, Springer, 2023, pp. 355–374.

J. He, C. Treude, and D. Lo, “LLM-Based Multi-Agent Systems for Software Engineering: Literature Review, Vision and the Road Ahead,” ACM Trans. Softw. Eng. Methodol., 2024, DOI: 10.1145/3712003.

X. Wang et al., “Openhands: An open platform for AI software developers as generalist agents,” in The 13th Int. Conf. on Learning Representations, 2024.

S. Hong et al., “MetaGPT: Meta programming for multi-agent collaborative framework,” arXiv preprint arXiv:2308.00352, vol. 3, no. 4, 2023, p. 6.

V. Dibia et al., “Autogen studio: A no-code developer tool for building and debugging multi-agent systems,” in Proc. 2024 Conf. on Empirical Methods in Natural Language Processing: System Demonstrations, Miami, FL, USA, ACL, 2024, pp. 72–79.

K. Zhang et al., “Codeagent: Enhancing code generation with tool-integrated agent systems for real-world repo-level coding challenges,” arXiv preprint arXiv:2401.07339, 2024.

M. Tufano et al., “AutoDev: Automated AI-driven development,” arXiv preprint arXiv:2403.08299, 2024.

W. Takerngsaksiri et al., “Human-In-the-Loop Software Development Agents,” arXiv preprint arXiv:2411.12924, 2024.

F. Zambonelli, N. R. Jennings, and M. Wooldridge, “Developing multiagent systems: The Gaia methodology,” ACM Trans. Softw. Eng. Methodol., vol. 12, no. 3, pp. 317–370, 2003.

S. A. DeLoach, M. F. Wood, and C. H. Sparkman, “Multiagent systems engineering,” Int. J. Softw. Eng. Knowl. Eng., vol. 11, no. 03, pp. 231–258, 2001.