learnxinyminutes: the repository where documentation is the source code

Documentation that has to compile in your head and in an interpreter

The project describes itself as code documentation written as code, which reads like a joke until you follow it through. The README's actual description is more careful: whirlwind tours of several, hopefully many someday, popular and ought-to-be-more-popular programming languages, presented as valid, commented code and explained as they go.

Three constraints in that sentence do the work. Valid means the code in a guide is real code, not pseudocode with the interesting parts omitted. Commented means the explanation lives in the comments rather than in prose paragraphs floating beside the snippet, so a reader who copies the block gets the explanation with the syntax. Explained as they go means the guides are ordered by difficulty rather than by reference completeness.

That combination is why the repository format is unusual. Every guide is a Markdown file containing a fenced code block in the target language, with section comments dividing it into the small number of headings that survive on one screen. Because the block is valid code, a reader can paste it straight into an interpreter and get output, which is a stronger check on a tutorial than any amount of prose review.

One Markdown file per language, plus translation directories

The root of the repository is a flat alphabetical list of language files. The visible listing runs from `active-oberon.md` through `agda.md`, `amd.md`, `ansible.md`, `bash.md`, `bf.md`, `c++.md`, `c.md`, `csharp.md`, `css.md`, `cmake.md`, `cobol.md`, `coffeescript.md`, `coq.md`, `crystal.md`, `dart.md`, and onward into `docker.md` and `dynamic-programming.md`. Along the way the naming shows the project's appetite: theoretical languages sit next to scripting ones, `bf.md` is Brainfuck, `bqn.md` is BQN, and `asymptotic-notation.md` and `dynamic-programming.md` are not languages at all but computer science topics filed the same way.

Some subjects are split across two files when they need it, which is visible in `clojure.md` sitting next to `clojure-macros.md`. So the unit of contribution is genuinely one file, and that is what makes the contribution instructions short.

Translations live in per-language directories rather than in suffixed filenames. The listing includes `ar/`, `be/`, `bg/`, `ca/`, `cs/`, `da/` and `de/`, which means Arabic, Belarusian, Bulgarian, Catalan, Czech, Danish and German versions of some guides, each directory holding the same filenames in its own language. The alphabetical run of the root listing is the clearest signal that the catalogue is far larger than any single directory view shows.

The contribution rule is copy an existing file and change the language

The README's recruitment pitch is the clearest statement of the workflow: we need YOU to write more inline code tutorials, just grab an existing file from this repo and copy the formatting, don't worry it's all very simple, then make a new file and send a pull request.

What makes this scale is that the format carries almost no decisions. There is no template engine, no site generator config in this repository, and no build step visible in the tree. A new guide is a Markdown file with the same comment-delimited structure as every other guide, and the rest is the language itself.

The process rules are equally specific, and two of them are practical rather than stylistic. Contributors are asked to prepend the tag `[language/lang-code]` to issues and pull requests, for example `[python/en]` for English Python, which is a routing convention that lets maintainers find the right subset. And anyone making more than one major change, with translations for two different languages given as the example, is asked to open a separate pull request for each one so that a reviewer can handle them individually. That is the whole review policy: smaller diffs, tagged by language, credited by name.

Licensing that lets a contributor walk away

The license section is unusual enough to be worth reading closely, and it is the one place in the repository where the terms are negotiated rather than imposed.

Contributors retain copyright to their work and can request removal at any time. By uploading a doc to the repository, the author agrees to publish it under the default Creative Commons Attribution-ShareAlike 3.0 Unported license, which is the license shown on each doc page. Anything not covered by that, which the README identifies as basically the README itself, can be used as the author sees fit, phrased as you can use as you wish, I guess.

So the guide text is copyleft with attribution, and the repository's own framing document is not. That distinction is unusual for an open source project and it explains something about the character of the catalogue: these are contributed documents first and a codebase second. The license field reads NOASSERTION rather than naming one grant for everything, which fits a per-file scheme better than a repository-wide one.

The attribution requirement also connects to the contributors instruction in the README, which tells you to fill in the contributors fields so you get credited properly. Credit and the share-alike obligation are two halves of the same deal here.

No releases, no versions, and a master branch that is the entire product

There is nothing to choose between when you arrive. The release list is empty, the default branch is `master`, and the last push was on 2026-09-21. What changed in that push is unknowable from the repository metadata alone, which is the honest characterization of a repository whose commits are all content changes and documentation fixes.

The surrounding files are equally spare. There is `CONTRIBUTING.md`, which the README links for the detailed style guide, a `LICENSE.txt` at the root, a `.mailmap`, `.gitattributes` and `.gitignore`. The homepage is learnxinyminutes.com rather than anything under github.io, which tells you the reading experience lives on a separate site and this repository is the content pipeline behind it.

The project has 12344 stars, 3683 forks and 242 open issues, and it is not archived. Those numbers read differently than they would for a library. Nobody is pinning a version, nobody is tracking a changelog, and forks are a reasonable way to take a guide and adapt it, which the licensing terms explicitly permit as long as attribution and share-alike are respected.

What a whirlwind tour can and cannot teach you

The name sets a low bar on purpose, and being honest about what that bar is worth more than inflating it. A guide in this repository will show you a language's comment syntax, its basic declarations, a few control structures, and often a distinctive feature that distinguishes it from its neighbours. In ten minutes you can learn enough to read the top of an unfamiliar file and recognise which constructs you are looking at.

What no guide here can do is teach the standard library, toolchain, package manager, or the idioms that experienced users of a language actually write. Those are exactly the things that make a language pleasant or painful to use, and they are the things a reference manual exists for. The project's own framing concedes the shape of the content: tours of popular and ought-to-be-more-popular languages, which is a selection about recognition rather than about depth.

There is also a research angle worth naming. A corpus of thousands of short, comparable, syntactically valid programs in different languages is an unusually clean dataset for anyone studying language design, and the consistent formatting rules make the files comparable in a way that arbitrary code samples from real projects never are. Nothing in the repository presents it that way, and nothing needs to for the corpus to be there.

Editorial conclusion

learnxinyminutes solves a problem no language vendor really wants to solve, which is giving a newcomer an accurate ten minute orientation to syntax that the official tutorial deliberately teaches slowly. The tradeoff is explicit in the repository's own framing: these are whirlwind tours presented as valid, commented code and explained as they go, useful as a map and not as a curriculum. What makes the project durable is not the individual guides but the shape of the contribution path, grab a file, copy the formatting, fill in the contributors fields, send a pull request. There are no releases to pick from, so `master` is the whole story, and the last push on 2026-09-21 was a content change rather than a versioned one. Start at learnxinyminutes.com for reading, and open a pull request against `CONTRIBUTING.md` if you want your language on the list.

Frequently asked questions

Official sources

1. adambard/learnxinyminutes-docs on GitHub
2. Issues
3. Project website
4. README