Agentic AI has become indispensable in software today. We use it ourselves, day to day, in the making of xprem: to reproduce an issue, explore the codebase, write tests, prototype features and write code.
So we are not opposed to AI-assisted contributions. What we are firmly opposed to is vibe coding, and AI slop in all its forms.
xprem ships an OTA update server that delivers code onto the devices of end users. It is a critical piece of infrastructure, and it calls for meticulous work and a real focus on security and performance.
We read and challenge each line of code carefully, whether a human or an AI wrote it, so that the quality bar stays high.
Using AI to contribute to this project is welcome, but you have to understand the code it produces, and read it carefully. We will very often challenge the code you submit. If you cannot defend your implementation, reviewing it will cost us more time than if you had opened an issue and let us implement it ourselves.
So before you contribute, take the time to understand the project and its architecture.
Issue and pull request bodies are communication between humans. Please do not ask an AI to produce six unreadable paragraphs for a small change. Say it plainly. We would rather read an incomplete pull request body than an unreadable pile of AI slop.