Why media entries need a different verification standard
A code tool can be checked by reading the repository. A video model cannot — so media verification leans on dated observation, and says so. Our…
Tag: <span>methodology</span>
How to evaluate a coding agent in thirty minutes
A repeatable test any team can run: five tasks, one checklist, and a scoring rule that tells you whether an agent fits your repository before you…
AEO: how to write a page an AI answer engine can quote
Answer engine optimisation is not keyword stuffing for chatbots. It is structure: a direct answer, extractable facts, and a visible date. Here is the pattern we…
GEO without link schemes: what actually gets a page cited
Generative engine optimisation has attracted the same tricks as SEO. The difference is that answer engines do not count links — they check whether a claim…
Why we label ‘official’ MCP servers separately — and what the label does not mean
An official server is maintained by the vendor. That is a maintenance signal, not a quality guarantee. Here is how the label works on this site…
We dropped 44 packages that were not MCP servers
For two weeks our package table had a defect we did not notice: it was sorted by downloads, and the top entry was Playwright… For two…
Why only 27% of our listings are reviewed
A directory has two levers: how many entries it has, and how much it can vouch for each one. Pulling the first is easy.… A directory…
What the verification date on a listing actually means
The date in the sidebar of every reviewed entry is the least interesting thing on the page and the most important thing about it.… The date…
What we refuse to publish
Every directory is defined by what it leaves out. Ours leaves out four things on purpose, and each exclusion costs us something. 1. Traffic… Every directory…
An npm keyword is not a directory
The naive way to build an MCP directory is to query npm for the keyword and publish what comes back. That produces a directory… The naive…