azl397985856's leetcode repository is CC BY-NC-ND, and its last tag is from 2022

CC BY-NC-ND blocks the two things most readers would want

The one machine-readable licence statement in this repository is in package.json, and it is the most restrictive content licence in common use.

{
  "scripts": {
    "book": "gitbook epub . && gitbook pdf . && gitbook mobi . "
  },
  "license": "CC BY-NC-ND 4.0"
}

CC BY-NC-ND 4.0 grants you the right to share the work unchanged and denies two other things. NonCommercial rules out using it in paid content or inside a product, and NoDerivatives rules out producing an adaptation, which in practice means a fork that rewrites the explanations is not permitted. LICENSE.txt sits in the root, and the repository metadata itself declares no licence at all, so a tool that reads repository metadata rather than package.json will report nothing. For a reader with no commercial use in mind that costs nothing. For a bootcamp, an internal handbook or a translated edition, it is a wall, and the README points to no alternative grant.

The newest tag is from 2022, the newest commit is from 2025

Release tags stopped more than three years before development did. The three newest are v0.37 on 2021-09-03, marked as the fifth cohort of the 91 day activity, v0.38 on 2021-09-04, marked as a book publication, and v0.39 on 2022-04-16, marked as the English version of the topic write-ups. The last push to master was on 2025-07-16 and the repository is not archived.

So the tag list and the commit list tell different stories, and only one of them describes the current state of the files. Anyone who installs or downloads by tag gets the files as they stood in 2022, and anyone who reads the star history or the commit list sees movement up to mid-2025. The contents are dated in the same way: the topic list in the README ends with a commented-out entry for a basic algorithms page, so a link that existed in an earlier version no longer resolves. Nothing in the repository states which platform problems were current when a given write-up was written.

Three of the five sections live outside the repository

The README divides the repository into five parts, and the tree only contains some of them. The classic problem analyses and the data structure summaries are files, under problems/ and thinkings/. The Anki cards, the daily problem and the plans are not.

The daily problem is described as an activity run in a chat group on WeChat and QQ, where a group solves one problem together, and where problems are later screened and added to the solutions section. The Anki cards record problems in a spaced repetition format, which means the format only does its work inside a particular application. The plans area records what will be added to the first three sections. None of the three is in the repository, and the same applies to the one-to-one tutoring, which is arranged over WeChat with a contact name the author hands out. Of five advertised sections, two are files you can read and three are services with an expiry date of their own.

The same Markdown has three publishing paths

The root carries the configuration for three different site builds and no statement about which one is current. There is book.json and a SUMMARY.md table of contents, which are GitBook files, and the online reading address given in the README is a GitBook URL. There is a _config.yml, which is Jekyll configuration, and a .nojekyll file, which tells GitHub Pages to skip Jekyll. There is a yarn.lock beside a package.json, though that package.json has one script. A .mdlrc and a .mdl_style.rb configure a Ruby markdown linter.

The consequence is that the book you read online is not necessarily built from the branch you are looking at. Regenerating it means running GitBook locally, and the README says the electronic edition is a limited-time free download that may be charged for later, with access handed out through the author's official account. So the convenient reading copy has a publication policy attached to it, and the tree keeps three sets of build files for one body of text.

The README doubles as a storefront for paid things

The same document that indexes the write-ups sells a book, a course, a membership and a plugin. The published book is linked through affiliate URLs on a Chinese retail site, in both physical and electronic form. LeetCode Premium is promoted with a personal referral link that adds two extra months to an annual membership and two extra weeks to a quarterly one, and the surrounding text is candid about why: the free problems already cover the classic cases and every problem type, but many companies' real interview questions stay locked. A 91 day paid activity has its own landing page and a discussion thread. One-to-one tutoring is priced by arrangement, and a browser extension is offered from the Chrome Web Store with an offline copy available by replying to the same official account.

None of that is hidden, and a reader can skip it. It matters because the offline copy of the extension is distributed outside the store, which means the version with automatic updates is not the version everyone ends up running.

Problem files are named with a padded four-digit number

The problems directory is navigable in a way most solution repositories are not. Filenames are zero-padded to four digits followed by the problem slug: 1.two-sum.md, 20.valid-parentheses.md, 21.merge-two-sorted-lists.md, 88.merge-sorted-array.md, 121.best-time-to-buy-and-sell-stock.md. Sorting the directory sorts by problem number. Some filenames carry a language suffix, as 53.maximum-sum-subarray-cn.md does, and entries in the easy list marked with 91 came out of the 91 day activity rather than from the author's own practice.

Two things follow. The count in the README, more than 200 write-ups and roughly 300 counting the topic problems, is a self-reported figure with no script in the repository that verifies it, and no file records which revision of a problem a solution was written against. And the topic index carries a disabled entry.

<!-- - [基础算法](./thinkings/basic-algorithm.md) -->

A commented-out list item is a page that was once linked and is now not, so a link shared from an older copy of the README will land nowhere.

The interview guide is a PDF the author found, not wrote

One of the most-used assets is third-party. The README points to assets/cheatsheet.pdf, an Interview Cheat Sheet the author describes as something found online, which opens with three standards for good code, readability, time complexity and space complexity, and then lists fifteen steps for an algorithm interview, beginning with writing down the key points once the interviewer has finished asking. The author's reaction to it is enthusiastic, and it is worth knowing that document did not originate here.

The same applies to the diagrams. A mind map of question types asked in internet company interviews is credited as an image from LeetCode, and the README carries a separate entry explaining how to fix images that fail to load, pointing at a separate proxy project. The written explanations survive a broken image host, the diagrams frequently do not, and anyone who redistributes this repository is redistributing a PDF and a set of images whose licences are not stated anywhere in the tree.

Editorial conclusion

This repository fits a reader who wants worked explanations in Chinese, organised by topic, and who reads the topic summaries before the individual problems. It does not fit an organisation that wants to reuse the write-ups, because CC BY-NC-ND 4.0 permits sharing but not adaptation and not commercial use. It also does not fit someone expecting a maintained problem set, since the newest tag is v0.39 from 2022-04-16 and the daily-problem and Anki cards sit in chat groups rather than in the tree. Read LICENSE.txt and package.json first, then weigh the last commit date, 2025-07-16, against how current the platform's problem list is for you.

Frequently asked questions

Official sources

1. Official documentation
2. Official README
3. Project repository
4. Release notes