How freeonnet is curated

A directory is only as good as what it refuses to list. This page is the rulebook.

Links only, never files

We point at official homepages. We do not host binaries, torrents, books, or models. If a project disappears, the link dies — we do not keep a mirror.

Legal and ad-safe

The main site only publishes resources that are legal to recommend and safe to put next to advertising: open-source tools, genuine free tiers, public APIs, public-domain media, and courses the authors meant to give away.

Grey-area streaming, cracks, and warez are out of scope here. The schema can route them to a separate site later; they will never appear on this one.

What the quality gate rejects

Newspaper op-eds filed as tools. Product-announcement blog posts instead of the product. Show HN sentences used as a name. README links like Issues or License. Descriptions that only repeat the title.

Whatever survives is renamed, classified by what it actually is (tool, library, course, list…), and scored. An LLM can repair a leftover; it cannot override a reject.

A small human overlay

For the resources people actually land on, we keep a short overlay: a tagline, a description, the real kind. Harvested copy is whatever the page said that day. The overlay is what we are willing to put our name on. We do not invent entries for things we do not know.

How ranking works

Editorial quality decides whether something is worth listing. Popularity (stars, votes) decides whether it belongs on the front page. A well-written 400-star Show HN is welcome in the catalogue and is not a featured pick.

Your visit stays here

Logos are fetched once by our pipeline and served from this site. Opening a page here does not tell Google, or the resources we list, that you looked — only clicking through does. No web fonts from third parties, no trackers.

The site also tells your browser so: its Content-Security-Policy only allows images, scripts and styles from this domain, so a third-party tracker could not load even if one were added by mistake.

Languages

The interface is Traditional Chinese, English, and Japanese. Resource names stay in the form the project uses; descriptions are translated when we have them, and fall back to English otherwise.