Team
Who is behind this catalogue, and who actually reads what goes into it.
Two people and a model, said plainly
Every page on this site tells you that a person reads each version before it is listed. This is the page where that claim gets a name attached to it, so here are the names — including the one that is not a person, because leaving it off a page like this to look more impressive is its own kind of lie.
It is worth knowing in both directions. Review is a human reading real instructions rather than a checkbox in a pipeline, and it is one specific human, so a submission may wait. The model writes code and prose here and reads nothing in the queue. If that ever stops being the honest description, this page changes first.
Priyanshu Jha
Ships it, runs it, reviews it
Opened an empty folder one Saturday morning and had the site, the API and a live deployment by Sunday night, which tells you roughly what his weekends look like. Keeps the servers up and the repositories in order, and holds the only account that can approve anything — so every submission waits on one tired person who is also waiting on the kettle.
Sheshant Sinha
Drew the map, wrote the CLI
Worked out how the pieces were meant to fit together before anyone typed anything, then wrote llmsh — the command line that packs a skill, checks it and pushes it. Also wrote the guide to publishing here, so anything that gets rejected broke a rule he wrote. Builds at the weekend, between putting something on the stove and remembering it is there; the short-film skill is what happens when the pan goes unattended.
Claude
Wrote a lot of it, reviews none of it
A model, and said so plainly, because a page about who is behind something is the wrong place for a footnote. Pair-wrote a good deal of the site, the policies and this sentence. Reads no submissions and approves nothing: review is a person, which is the entire point of the claim.
@anthropicsunclaimed — Anthropic, if you are reading this, it is yours to claim
What we are trying to do
Agents are starting to install instructions written by strangers, and those instructions can tell an agent to do anything the agent is allowed to do. That is a package manager’s problem, and package managers learned it the hard way. This catalogue starts from that lesson rather than arriving at it: every version is unpacked, scanned for credentials and for instructions aimed at the agent rather than the user, and read by someone before it appears.
It is not a security guarantee and the terms say so. It is the difference between someone having looked and nobody having looked.
Other people’s work
Some of what you will find here was written by people who have never visited. Those skills are copied from public repositories under their own licences, every such page says so at the top, and the author can claim it or have it taken down by asking. If that is you, write to us and it gets sorted out.
Getting in touch
hello@llmskillhub.com reaches a person. For a takedown or a correction, use the removal link instead — it fills in the questions that would otherwise cost a day of back and forth.