juunge

RALPH.md is a skill-like format for reusable agent loops. It defines the outer loop around an agent run: the agent command, the commands to run between iterations, the arguments for steering, and the prompt body. That makes it possible to package and share loops that run repeatedly until stopped. I’ve been writing down example patterns and use cases in the cookbook as I learn what works well: https://ralphify.co/docs/cookbook/

That would be great! And thanks for your take on this problem, I love to see others take on solving it

I really like the idea of an agent package manager! I built something similar aiming to be like uv but for AI agents https://github.com/kasperjunge/agent-resources

I’ve made a CLI tool that does something similar, called Copcon:

https://github.com/kasperjunge/copcon

Point it at a code project directory to get a file tree and content, optionally with a git diff, copied to the clipboard - ready for copy pasting into ChatGPT.

It is very true that this only works for small projects, as you will bloat the LLM’s context with large codebases.

My solution to this is two files you can use to steer the tool’s behavior:

- .copconignore: For ignoring specific files and directories.

- .copcontarget: For targeting specific files and directories (applied before .copconignore).

These two files provide great control over what to include and exclude in the copied context.

Awesome reverse engineering project. Next up: reverse engineer a NN's solution to Fourier Transform!

I was very surprised about the "talk to your customers" part