GD-KS: A Spec-Driven Multi-Agent Framework for LLM-Assisted Game Development
Resumo
Introduction: LLM-based coding assistants operate mainly at the level of code completion and do not model the process by which a software artifact is produced. This gap is acute in game development, whose pre-production (concept, design, narrative, art, audio) generates documentation whose volume and interdependencies rival the implementation effort — a burden falling disproportionately on indie studios and solo developers. Objective: To design a multi-agent framework for indie and solo developers that applies SpecDriven Development (SDD) to the game development lifecycle and decouples engine-agnostic design from engine-specific implementation. Steps: We designed GD-KS, an open-source four-phase spec-driven pipeline (ideation, design, planning, engine) with up to 32 specialized agents. Its main technical innovation is a plugin-based engine-profile abstraction isolating engine-specific knowledge from design artifacts, currently supporting Unreal Engine 5, Godot 4 and Unity 6, complemented by a single-source-of-truth project state with append-only audit trail. The framework is validated through schema validation, 240 automated tests, and multi-engine installation scenarios. Expected Results: The pipeline enforces documentation discipline that small teams often cannot maintain manually; the engine-profile abstraction lowers the cost of evolving to new engines; the project state supports fragmented sessions typical of solo work. A user study is planned as the next step.
Referências
Anthropic (2024). Claude Code: Agentic Coding in the Terminal. [link]. Accessed: 2026-07-26.
Anysphere (2024). Cursor — The AI Code Editor. [link]. Product documentation. Accessed: 2026-07-26.
BMAD (2024). BMAD-METHOD: Breakthrough method of agile AI-driven development. [link]. Accessed: 2026-07-26.
Chen, D., Zhang, H., Wang, H., Huo, Y., Li, Y., e Wang, J. (2025). GameGPT: Multi-agent collaborative framework for game development.
Codecks, Inc. (2015). Codecks: Project management for game developers. [link]. Accessed: 2026-07-26.
CrewAI Inc. (2024). CrewAI: Framework for orchestrating role-playing, autonomous AI agents. [link]. Accessed: 2026-07-26.
Gallotta, R., Todd, G., Zammit, M., Earle, S., Liapis, A., Togelius, J., e Yannakakis, G. N. (2024). Large language models and games: A survey and roadmap. IEEE Transactions on Games, pages 1–18.
GitHub (2024). GitHub Copilot — Your AI Pair Programmer. [link]. Accessed: 2026-07-26.
GitHub, Inc. (2025). GitHub Spec-Kit. [link]. Accessed: 2026-07-26.
Grapken Contributors (2024). Grapken: Visual node-based editor for indie game design docs. [link]. Accessed: 2026-07-26.
HacknPlan Ltd. (2017). HacknPlan. [link]. Accessed: 2026-07-26.
Hassan, A. (2025). Automated Unity game template generation from GDDs via NLP and multi-modal LLMs.
Hong, S., Zheng, X., Chen, J., Cheng, Y., Wang, J., Zhang, C., Wang, Z., Yau, S. K. S., Lin, Z., Zhou, L., et al. (2023). MetaGPT: Meta programming for multi-agent collaborative framework. arXiv preprint arXiv:2308.00352.
Hou, X., Zhao, Y., Liu, Y., Yang, Z., Wang, K., Li, L., Luo, X., Lo, D., Grundy, J., e Wang, H. (2024). Large language models for software engineering: A systematic literature review. ACM Transactions on Software Engineering and Methodology, 33(8):1–79.
Hu, S., Huang, T., Liu, G., Kompella, R. R., Ilhan, F., Tekin, S. F., Xu, Y., Yahn, Z., e Liu, L. (2025). A survey on large language model-based game agents.
Keith, C. (2010). Agile Game Development with Scrum. Addison-Wesley Professional.
Li, G., Hammoud, H., Itani, H., Khizbullin, D., e Ghanem, B. (2023). CAMEL: Communicative agents for “mind” exploration of large language model society. arXiv preprint arXiv:2303.17760.
Nuclino, Inc. (2016). Nuclino: Unified workspace for game design documentation. [link]. Accessed: 2026-07-26.
OpenDevin Contributors (2024). OpenDevin. [link]. Accessed: 2026-07-26.
OpenSpec Contributors (2025). OpenSpec. [link]. Accessed: 2026-07-26.
Pezoa, F., Reutter, J. L., Suarez, F., Ugarte, M., e Vrgoč, D. (2016). Foundations of JSON schema. In Proceedings of the 25th International Conference on World Wide Web (WWW), pages 263–273. ACM.
Piskala, D. B. (2026). Spec-driven development: From code to contract in the age of AI coding assistants.
Qian, C., Cong, X., Yang, C., Chen, W., Su, Y., Xu, J., Liu, Z., e Sun, M. (2023). ChatDev: Communicative agents for software development. arXiv preprint arXiv:2307.07924.
Schell, J. (2019). The Art of Game Design: A Book of Lenses. CRC Press, 3rd edition.
Serra, C. B., Serra, G. M. B., e de Classe, T. M. (2025). Digital game development using large language models (LLMs): An exploratory study. In Anais do XXIV Simpósio Brasileiro de Jogos e Entretenimento Digital, pages 538–549, Porto Alegre, RS, Brasil. SBC.
Silvestrini, M. M. (2026a). GD-KS: Game development knowledge system. [link]. Open-source repository, version v0.4.1-beta.2. Accessed: 2026-07-26.
Silvestrini, M. M. (2026b). gd-ks npm package. [link]. Version 0.4.1-beta.2. Accessed: 2026-07-26.
SoloMode, Inc. (2026). SoloMode: All-in-one tool for solo game devs. [link]. Accessed: 2026-07-26.
StitionAI and Contributors (2024). Devika. [link]. Accessed: 2026-07-26.
Tang, S., Zhao, K., Wang, L., Li, Y., Liu, X., Zou, J., Wang, Q., e Chu, X. (2025). UnrealLLM: Towards highly controllable and interactable 3D scene generation by LLM-powered procedural content generation. In Findings of the Association for Computational Linguistics: ACL 2025, pages 19417–19435, Vienna, Austria. Association for Computational Linguistics.
Tessl, Inc. (2025). Tessl. [link]. Accessed: 2026-07-26.
Wu, Q., Bansal, G., Zhang, J., Wu, Y., Li, B., Zhu, E., Jiang, L., Zhang, X., Zhang, S., Liu, J., Awadallah, A. H., White, R. W., Burger, D., e Wang, C. (2023). AutoGen: Enabling next-gen LLM applications via multi-agent conversation. arXiv preprint arXiv:2308.08155.
Yin, L., Cheng, W., Qin, Z., Huang, T., Li, Y., e Ding, G. (2026). AutoUE: Automated generation of 3D games in unreal engine via multi-agent systems. In Findings of the Association for Computational Linguistics: ACL 2026. Association for Computational Linguistics.
