§ 02 — the answer, in one example
This is what a linux-meta entry answers.
What it is, in plain words · for which age · and where each fact came from. No technical English, no guessing.
§ 03 — start from a need
Ready-made lists and real categories.
Parents pick by age, teachers by subject, creators by media. Or browse the whole catalog by category.
For kids (6–9)
Educational apps with a checked age.
- GCompris 0+
- Tux Paint 0+
- KTouch 0+
- Kanagram 0+
For schools
Science, maths and productivity.
- GeoGebra 0+
- Stellarium 13+
- Kalzium 0+
- KGeography 0+
Creative & media
Free image, audio and video.
- Blender 0+
- Krita 0+
- GNU Image Manipulation Program 0+
- Kdenlive 0+
Free alternatives
Open swaps for proprietary apps.
- Thunderbird 7+
- Inkscape 0+
- GNU Image Manipulation Program 0+
- darktable 0+
§ 04 — why you can trust the entry
Every fact says where it came from.
That is what makes the age and the description trustworthy: nothing is presented as fact without an origin. Current coverage, in the open.
- imported — redistributed from upstream
- AI — suggested by workers, awaiting review
- human — reviewed by an identified contributor
| Category | Total | Detail |
|---|---|---|
| unique software | 77,221 | 80,458 packages |
| age ratings | 2,107 | 717 imported 1,390 AI 0 human |
| packages with translated description | 89,952 | 2,836 imported 12,633 AI 74,483 human 389,743 (package × locale) pairs total · 4.3 locales/package avg |
| semantic embeddings | 80,453 | used for semantic search and grouping |
§ 05 — how the data is built
An open pipeline. No black box.
> ingest manjaro flathub aur debian → 80,458 indexed packages
> classify --oars --ai --review → 2,107 ratings with visible origin
> translate --locales pt,en,es,de,… → 89,952 packages with translated descriptions
> publish --html --dataset --appstream → /en/p/firefox- 01
Ingest
Workers collect metadata from Manjaro, Flathub, AUR and Debian. Then we dedupe by canonical_slug: Firefox remains Firefox on any distro.
- 02
Classify
When upstream OARS exists, we preserve it. Where there is a gap, workers suggest a rating and human review can confirm or correct it.
- 03
Publish
The site shows the data with origin and the full dataset enables local integration. People decide better; distributions integrate with less manual work.
§ 06 — why this matters now
Clear metadata becomes a base, not guesswork.
The world is standardizing software metadata with clearer age, risk and origin. linux-meta is open infrastructure for that — not legal advice.
- EU · Digital Services Act
- UK · Online Safety Act
- Brazil · ECA Digital
- Australia · minimum age
§ 07 — how you take part
The next reviewed rating can prevent a bad decision.
You do not need to maintain a distro to help. Searching, reporting an error, reviewing one rating or translating one description already improves data other people will reuse.
- 01 · No sign-up
Look up a program
See whether the page helps you decide. Found something confusing, wrong or incomplete? That feedback matters.
Open catalog → - 02 · Free account
Review one data point
One reviewed rating, translated description or corrected origin already improves the dataset for everyone.
Create account → - 03 · Coming
Sustain coverage
Responsible sponsorship needs accountability. The financial channel only opens when transparency is ready.
Express interest →
Building a distro, store or research? Reuse the open dataset (JSON · CSV · AppStream) →
§ 08 — frequently asked
Short answers to important questions.
- Who decides the age?
- We first preserve OARS ratings from upstream. Where OARS is missing, the project can suggest a rating and keep the origin visible until human review.
- How do I use this in my distro?
- Use the full snapshot when it is available on the transparency page. The goal is local consumption without depending on a linux-meta API.
- Are you neutral?
- The goal is traceability. Each field should show origin, method and review state so anyone can contest it.
- How is money used?
- The financial channel still depends on public accountability. Until then, the most useful contribution is reviewing, translating, testing and reporting errors.