What this site is

AI Agent Hub (ai-develop.top) is an English-language directory. It collects public listings of agent skills, DeepSeek Harness plugins, and large language models, and it publishes a separate blog of engineering articles. The directory is here so you can find a name, a publisher, an install prompt, and a link back to the original project. It is not a claim that we built those projects.

What we write ourselves

Three kinds of pages are written for this site and are not copied from a publisher:

  • Guides explain how to choose, install, and reject a skill or plugin, and how to read a model comparison.
  • Category introductions explain what a section of the directory is for, who it is a bad fit for, and how we sort it.
  • Editor's notes appear on a small set of individual listings. Each note says what problem that item is for, who should skip it, the first three steps, and where else on this site to look. A listing without that note is still a directory card.

Where listings come from

Skill and plugin entries point at a public source URL or a GitHub repository. Model entries carry the description, license, context window, and other fields published by the model vendor or a public model card. We keep those links so you can check the upstream page. Descriptions that arrived with the listing are the publisher's material or a summary of it. They are not an independent review.

How ordering works

Skill lists default to a score, then downloads. Plugin lists default to GitHub stars. Model lists default to the directory score. A high score means "sorted near the top of this directory," not "safe to install" and not "best model." The guide How ranking works says what each number is and what it is not.

Who operates it

The site is operated independently as AI Agent Hub. We are not Tencent, DeepSeek, Anthropic, OpenAI, Google, or the publisher of any listing, unless a listing's own page says otherwise. To correct a fact, request a takedown, or ask a question, use the contact form. Messages are stored so we can reply. The privacy policy and terms describe what that means.