9 November 2026
0 of 302: what a full census of the MCP Registry says about the 2026-07-28 revision
A seeded random sample of packaged servers, a Wilson interval around a zero, and the 18 packages that could change it.
Writing
Each post gives the method, the result with its caveats, what it does not show, and how to reproduce it.
9 November 2026
Twelve ways an AI-built Supabase + Stripe store leaks customer data, each with a proof you can rerun
Four root causes behind twelve planted flaws, ten of them proven in CI on every push.
9 November 2026
I deleted a census tier I’d already built: scope files versus random sampling
Why a random sample of strangers’ servers can’t have an authorising party, and what the census lost by keeping the rule.