Yup. It’s a system that worked when it was relatively safe to assume a chain of trust. But it’s not scaling to the era of AI agents writing patches, nor the increasingly number of attacks against existing foundational packages.

We really do need to rethink the security model behind open source.