agentmods.dev

Terms

Two licences live on this site, and the line between them is the whole point of this page. The catalogue — the records, the measurements, the links between them — is offered under the Open Database License. The file bodies it describes are their authors' and stay under whatever licence their repository carries. Neither reaches the other.

The catalogue: ODbL 1.0

The database behind this site is made available under the Open Database License, version 1.0. That covers every record the crawler produced: the name, type, path and repository of each mod; the measurements — tokens per session, tokens once invoked, scan grade and findings, origin and similarity, stars and their daily change; the plain-language descriptions; the categories; and the relations between records, such as which plugin ships which skill and which copy descends from which original.

In plain terms the licence asks two things of you:

  • Attribute. Where you show or redistribute data from here, say where it came from: Data: agentmods.dev, ODbL 1.0, with a link. In an API response, a footnote, a README — anywhere a reader of yours could find it.
  • Share alike. If you publish a database that is derived from this one — a copy, a filtered copy, a copy with your own columns added — it must be offered under ODbL too. Using the data inside a product, a tool or an agent is not publishing a database and carries no such obligation; publishing the dataset is.

The short version above is a summary and not the licence. The licence is the linked text, and where the two differ, the text wins.

The bodies are not the catalogue

A mod's text — the SKILL.md, the rule, the agent file, the JSON — is not ours and is not under ODbL. It belongs to the repository it came from and stays under that repository's licence. Where that licence permits copying (MIT, Apache-2.0, BSD, ISC, MPL-2.0, CC0, Unlicense), the body is reproduced on the mod's page and served by the install command; where it does not, or where there is no LICENSE file at all, the page carries the record and a link and never the text. The rules for that are on the about page.

So: the fact that a file exists, where it is, what it costs and what the scan found are ODbL. The file itself is whatever its author said it is.

The API

The endpoints under /api/ — search, items, install, the remote MCP server, the update check — are free to use, without a key, for what they are for: an agent looking up a mod, a CLI installing one, a tool checking whether an installed one has changed. That is the distribution channel that lives inside the tools rather than on this site, and it is meant to be used.

Three rules keep it usable:

  • Attribute what you show. Data from the API that reaches a reader carries the attribution above. A response header, X-Data-Licence, restates it on every API answer so a client can pass it on.
  • Rate limits. Without a key, sixty requests a minute per address, which is far more than any interactive use needs and far less than a bulk pull. A response over the limit is 429 with Retry-After, and every response carries X-RateLimit-Limit, X-RateLimit-Remaining and X-RateLimit-Reset so a client can pace itself. Limits are counted in memory on the server and start over when it restarts.
  • No bulk redistribution without asking. Walking the whole catalogue through the API to republish it, or to build a competing catalogue from it, is what the share-alike clause and the rate limit are both there for. The way to do it properly is the section below.

Bulk and commercial access

If you want the dataset rather than lookups — to feed an IDE, a security tool, research, or an agent of your own — the honest path is a licence rather than a scrape. An API key lifts the rate limit; a data licence covers commercial redistribution beyond what ODbL allows. Both are granted by hand at the moment, because nobody has asked yet and a form nobody needs is a form nobody maintains.

Ask by opening an issue at github.com/iOSDevSK/agentmods and saying what you are building and roughly how much you would pull. There is no wrong way to ask.

Your repository

If your repository is listed and you would rather it were not, say so in the same place and it is removed and not crawled again. The about page says how the crawler behaves. The privacy page says what the site collects from you.

Changes

This page is a file in the site's repository and changes to it are commits, dated and readable. Last changed 4 September 2026.