TLDR on how it works: It's a small stack of skill files and some JS code that starts a large number of Codex sessions. They all get the prompt and the same scop

TLDR on how it works: It's a small stack of skill files and some JS code that starts a large number of Codex sessions. They all get the prompt and the same scope with a limited set of tools. Not one prompt in the repo contains security advice, performance is obtained only through scaling the number of agents looking at the code