Hey fellow devs, I’ve been exploring Agentic Coding — the idea of giving AI agents goals rather than tasks — and I’m curious if anyone here has applied this to actual game development, either in tooling or gameplay. Think Claude, GPT, or similar LLMs acting less like autocomplete and more like a junior engineer that plans, reasons, tests, and even iterates based on broader goals you give it. I’m especially interested in: Resources, tools, or frameworks built around this approach Examples of devs using LLMs as autonomous assistants in codebases Ideas for games that simulate or incorporate agentic thinking (e.g., AI NPCs, empire managers, agents with goal hierarchies) Thoughts on integrating this into build pipelines, editors, or scripting languages I’m working on a sci-fi 4X game called Stellar Throne, and I’m experimenting with agentic workflows using Claude and ChatGPT (I’ve even built a custom CLAUDE.md spec to guide them). Would love to compare notes or find others deep in this rabbit hole. I’ll compile anything useful from this thread into a blog post. Thanks in advance! – MrPhil mrphilgames.com