Today, AI agents are static - they're built with a fixed set of tools, prompts, and configurations that remain unchanged throughout their operation. But what if an agent could evolve and adapt its own capabilities based on the tasks it encounters?

This came from frustration with the agents I’ve been using to automate my workflows—they keep breaking on edge cases. Soon, I want it to take the day’s conversations, build a fine-tuning environment, and train itself overnight so it’s better for me the next day.

This was a fun hackathon project, but I'd love to continue hacking on it with others! Feel free to DM me if you're interested.

Github: https://github.com/tejpalv/god-agent/tree/main