smol developer

a fork of swyx's smol developer. Trying out new ways of extending it.

running

pip install modal-client
modal token new

modal run main.py --prompt flask.md

modal run debugger.py --prompt flask.md

if you don't set a concurrency_limit on calling the api you get rate limited by openai. It works with

n = 10 for the flask prompt and takes 2 mins,
n = 5 and takes ~4 mins
n = 20 also took ~2 mins so I think at that point open ai's API becomes the rate limiting factor