Hi everyone, I'm new to AgentGPT and have been experimenting with creating a few different agents for project management workflows. I'm coming from a background where I've used tools like Jira and Linear extensively, so I'm used to having robust versioning for my project configurations and workflows.
As I build out my AgentGPT agents, I'm realizing that the prompts, goals, and constraints I set are essentially the "source code" for these assistants. I've already tweaked one agent a few times and wish I could easily revert to a previous version or compare changes. I don't see a built-in version history feature within the AgentGPT interface.
Could anyone share their practical methods for managing this? Specifically, I'm curious about:
- Are you simply copying and pasting the agent configuration into a separate document or tool?
- If you use a text file or a markdown document, what's your naming and folder structure to keep track of iterations?
- Has anyone tried using a proper version control system like Git for their agent configs? If so, how do you structure the repository—one file per agent, or a more complex setup?
- Are there any tools that integrate with or complement AgentGPT to handle this aspect?
I want to avoid losing a good working configuration by making a series of small changes that might not pan out. Any detailed advice or comparisons between different approaches would be very helpful.
Thanks!