RegexLearn: a regex course that ships 21 translations and a private playground

Two lesson tracks rather than one long tutorial

The README describes the platform as turning regular expressions from a black art into a skill you can pick up in an afternoon, and the curriculum is built around two tracks rather than one exhaustive course.

The first is Regex 101, covering the fundamentals. The second is Regex for SEO, described as practical, search-focused usage, which is an unusual and useful framing: it teaches regex for the job people most often need it for instead of stopping at theory. Both are interactive lessons that progress from basics to advanced patterns one small step at a time, and the whole thing is keyboard navigable, which matters more than it sounds for a course you are meant to finish in one sitting.

The two-track structure is the design decision that makes this different from a tutorial. A single linear course has to pick a pace and an audience. Splitting fundamentals from an applied specialisation lets the SEO track assume you can already read a character class, which shortens it considerably.

The playground is private, which is the correct default

The feature list names four things: step-by-step lessons, the two courses, a cheatsheet, and a playground described as a private, browser-based sandbox to write and test regex patterns locally.

Private is doing real work in that sentence. Regex testers have a long-standing habit of shipping your pattern, and often your test string, to a server so someone can rank the pattern in their database. If you are testing a pattern against customer data, an internal log line, or a credential format, that upload is a disclosure you did not agree to. A playground that runs entirely in the browser has no such problem, and this one says so explicitly.

The cheatsheet fills the other gap these tools usually leave. A playground is good for exploring, bad for recalling. Being able to glance at a syntax summary while you work is what turns knowing a concept into being able to use it.

Adding a language is a folder of JSON and a pull request

The supported language count is 21, and the README treats that as an invitation rather than a boast. The instruction is specific: copy the `en` folder in `src/localization/`, translate the JSON files, and open a pull request.

That is a genuinely low bar and it is the sort of thing that determines whether a project gets translated at all. Most localisation efforts fail on review load, not on translation difficulty, and a workflow where the reviewer is checking a mechanical copy of an existing structure is one people actually do.

The languages listed are English, Turkish, Russian, Spanish, Simplified and Traditional Chinese, German, Ukrainian, French, Polish, Korean, Brazilian Portuguese, Czech, Georgian, Persian, Italian, Arabic, Bengali, Japanese, Indonesian and Vietnamese. The recent release history matches the pattern: v2.36.0 on 2026-07-15 added Japanese, Indonesian and Vietnamese in a single day, and v2.35.0 earlier that day added Bengali alongside completing Traditional Chinese.

Running the site locally

The stack is Next.js, TypeScript and Tailwind CSS, and the development instructions are three commands:

npm install   # install dependencies
npm run dev   # start the dev server at http://localhost:3003
npm run build # production build

Port 3003 is the detail to note if you have several Next.js projects open, since the default 3000 is often taken.

The rest of the scripts come from `package.json`. `next build` is the production build, and a `postbuild` hook runs `next-sitemap` followed by a local script that fixes the HTML language attributes, which is the kind of thing you need when the site serves 21 locales and one page template. `npm run lint` is `eslint . --ext .js,.jsx,.ts,.tsx`.

Dependencies include `react-intl` for the message catalogue, `draft-js` for the lesson editor, `headlessui` for components and `canvas-confetti`, which tells you the lessons celebrate completion. Local environment variables belong in `.env` or a `.env.*.local` file, and the README is explicit that those are gitignored and must not hold values meant for source control.

Branch names, releases and who pays for it

There is a contradiction in the contributing guide worth knowing before you open a pull request. The README asks you to target the `develop` branch and calls it the default and integration branch for upcoming releases, while stating that `main` is production and deploys to regexlearn.com automatically. The repository metadata for the default branch says `main`. So the guidance is written for a flow where work lands on `develop` and is promoted, and the hosting default simply points at the production branch.

Release cadence is fast and translation-led. Recent tags include v2.36.0, v2.35.0 and v2.34.3, all published on 2026-07-15. Earlier work shows the same pattern of small fixes: read-only regex results rendering correctly on load, answer hint positioning on mobile, and a fix to Playground end-of-line anchor matching. v2.35.0 also removed the retired Playground sharing API along with shared-link loading, which is a privacy improvement and a reminder that an earlier version of this tool did send patterns somewhere.

The project is MIT licensed and free, funded by sponsorship rather than a paid tier. Ahrefs and Wope are the listed sponsors, and the sponsorship pitch puts your logo in front of developers at the moment they are learning.

Editorial conclusion

RegexLearn sits in a specific gap: it is structured practice rather than a reference, which most of what Google returns for regex is not. If you learn by being walked through a progression and you can read one of the 21 languages, it will get you from reluctant to competent in an afternoon. If you already write regex daily and want a fast lookup, the cheatsheet alone is the part worth keeping. Start with the Regex 101 track rather than the playground, because the playground teaches you nothing on its own, and notice that the repository asks for pull requests against a `develop` branch while the repository metadata names `main` as the default. The last push was on 2026-07-15 and v2.36.0 shipped the same day with three new translations.

Frequently asked questions

Official sources

1. aykutkardas/regexlearn.com on GitHub
2. License: MIT
3. Project website
4. README
5. Releases