Ask HN: What rules does your team have for using AI in a durable codebase?
What guidelines/manifest do you have set up in your team for using AI responsibly to generate production code that has to be maintained long-term? Having an ADR and opinionated policies in our Django codebase has helped maintain quality despite extensive use of LLMs to generate code. It also helps that more effort has shifted to code review and we also use AI tools to help review PRs.