Harnesses, Playbooks, and Proof

In February, OpenAI published Harness engineering: leveraging Codex in an agent-first world. Ryan Lopopolo and his team spent five months building a product where no human wrote a line of code, and sh