Agentic coding increasingly removes humans from the process of picking libraries. To me it always had some "human" touch - you'd look at the library, core maintainers, follow them on social media, etc. But with AI the typical customer of an open source project is now an AI agent. Essentially making maintainers write code for some LLM model rather than another human being.

We've seen stories of a flood of PRs in popular repos, Tailwind having to cut staff despite being used by every agent, etc. It feels like it creates some disbalance in already tough market, and the change will only accelerate from here.

Any thoughts on how automation of software engineering will shape open source?