A note on AI usage¶

jero's design and architecture are entirely the author's — motivated by years of building on Django/DRF, Flask, Express, and FastAPI, and re-implementing the same patterns jero now formalises. The implementation and documentation (with the exception of this note), however, were heavily assisted by AI (OpenAI Codex).

The approach has been an informal spec-driven loop: a spec, usually with a small snippet of the desired user-facing interface, which the agent builds under a short human/agent feedback cycle. Tests carry extra weight in this model, so jero's suite is purposefully written to target only the public, user-facing interface: when the comprehensive suite passes, the user-facing behaviour is correct, and the AI is allowed freedom to refactor within those confines.

The author has endeavoured to proofread all the documentation and most of the generated code; no doubt some pieces remain that he himself might raise an eyebrow at. That, in the author's opinion, is one of the tradeoffs of fully embracing AI-assisted development.

If you disagree or have concerns, the author would love to hear your thoughts — please reach out on GitHub Discussions.