LeetCode-Master is a reading order, not a codebase, and its own index reuses the same problems twice

The whole repository is four entries, and nothing in it runs

The top level contains .gitignore, README.md, pics/, and problems/. That is the entire tree. There is no source directory, no package manifest, no test suite, and no build step, and each problem is a Markdown file under problems/ named after its number, for example 0704.二分查找.md for binary search, 0206.翻转链表.md for reversing a linked list, and 0015.三数之和.md for three sum. The recorded primary language for the repository is Shell, which is a property of the file mix rather than a toolchain anyone can invoke.

Nothing here installs or runs. There is nothing to install, nothing to run, and no test command to fail, so the usual way of evaluating a study repository, cloning and trying an example, does not apply. What a clone buys is the ordered set of write-ups and the images in pics/. The repository also has no GitHub releases, so there is no versioned artifact to pin, and the last push landed on 2026-08-03, which is the date to judge how current the material is. Anything you intend to rely on across months should be copied locally rather than tracked against a moving branch.

The two pointer chapter re-solves problems listed three chapters earlier

The stated selling point is that you work through the list in order, which means the same problem can appear more than once. The 双指针法 (two pointer) section is the clearest case: its eleven entries are 27.移除元素, 344.反转字符串, 替换数字, 151.翻转字符串里的单词, 206.翻转链表, 19.删除链表的倒数第N个结点, 链表相交, 142.环形链表, 15.三数之和 and 18.四数之和, and every one of them already appears in the 数组, 字符串 or 链表 chapter. The same pattern shows up elsewhere, with 1002.查找常用字符 filed under hash tables and 347.返回数组中的前K个高频元素 standing in for the same idea in the stack and queue chapter.

A technique gets a second pass from a different angle, which is the point of the repetition. The cost lands on scheduling, because the number of entries in the readme overstates how many distinct problems exist and a study plan counted from it comes out shorter than the work. Treat the technique chapters as a second visit to problems you have already met, and go back to them in the order given rather than skipping ahead on the assumption that each line is new.

Numbering has holes and the readme admits it shows only part of each chapter

Two signals tell you the directory in the readme is an index rather than a manifest. First, the numbering skips: the 字符串 chapter runs 1, 2, 3, 4, 5, 6, then jumps to 8, and the 二叉树 chapter runs through 14 and then jumps to 16. Entries were removed or renumbered at some point and the sequence was left with gaps, so a missing number does not mean a problem was never covered. Second, the note above the directory says the list is ordered for the learning curve and that only the first several problems of each chapter are shown, with the full list needing to be expanded. In the rendered page that expansion is a details block per chapter, and the binary tree chapter additionally carries an outline image rather than text.

Chapter length cannot be estimated from that list. If you are planning weeks of study, count the files in the corresponding problems/ directory instead, and expect the weekly recap entries under a 周总结 path, dated 20200927, 20201003 and 20201010, to appear in the middle of the binary tree list as numbered items rather than as separate review material. Those dated files are the review step the usage notes tell you to do at the end of a module, placed where you will meet them while working.

The first module opens with AI tool links before any complexity analysis

The 前序 chapter, which is where the reading order starts, has two groups and only one of them is about algorithms. The AI工具 group points at two sibling repositories: one for topping up Alipay and WeChat accounts for ChatGPT and Claude, and one titled for wiring up Claude Code Opus 5 and GPT5.6 API access. The 算法性能分析 group is the part with lasting value, and it is more concrete than a typical preamble: an article on time complexity, one on why an O(n) algorithm times out and how large n has to be for that to happen, one deriving the time complexity of recursion through a single interview problem, one on space complexity, one on the time and space cost of recursion, and one on what your code actually costs in memory.

Someone opening the syllabus for the fundamentals therefore meets two unrelated tool links before the complexity material that justifies the whole exercise. Within that group the O(n) timeout piece is the one that changes behaviour rather than knowledge, since it connects a written complexity to a judge's verdict, and it is worth reading before you treat a passing solution as a good one.

A handful of problems come from KamaCoder, not from LeetCode

Four entries in the array and string chapters live under a problems/kamacoder/ path and use that site's numbering instead of a LeetCode number: 0058.区间和 (range sum) and 0044.开发商购买土地 (the developer land purchase problem) in the array chapter, and 0054.替换数字 (replace digits) and 0055.右旋字符串 (rotate a string to the right) in the string chapter. The 字符串 chapter also reuses 替换数字 in the two pointer section, where it keeps the kamacoder path.

These four are not self contained inside the repository in the way the numbered LeetCode files are, since the statement each one solves is hosted elsewhere, and the Kamacoder notes link in the readme points to notes.kamacoder.com. A reader working strictly from a clone of this repository can read the explanation but may not be able to see the input format, which is the part those exercises usually hinge on. If you intend to work through the array and string chapters start to finish, resolve those statements on the other site first, or accept four exercises described rather than fully specified.

The English edition is a different repository, and the mirrors can be behind

The banner at the top of the readme points to four places to read this material, and they are not interchangeable. There is an overseas English edition at keetcoder.com, a separate English repository at github.com/youngyangyang04/keetcoder, the Chinese site for online reading at programmercarl.com, and a Gitee sync. Everything else in the file, including the chapter titles, the problem names and the theory sections, is in Chinese, so an English reader who clones this repository and expects translated write-ups will find the index and nothing more. The English text is the other repository's job.

Two hosts and one default branch named master means the same content is distributed from more than one place, and a clone taken from the Gitee sync can lag the GitHub copy if the sync is not continuous. The document itself, the published book, the algorithm PDF at programmercarl.com/qita/algo_pdf.html, the 170 episode video course on Bilibili, and the interview notes at notes.kamacoder.com, all live outside the repository, so the clone is the index and the lessons are somewhere else. The contribution guide is also external, at programmercarl.com/qita/join.html, which is where to go before sending a pull request.

No LICENSE file, no releases, and a reuse rule phrased as a warning

No licence is declared anywhere in the tree, which holds only .gitignore, README.md, pics/ and problems/, and there is no LICENSE file to open. What the readme offers instead is a notice stating that all of it is original work, that quotations should credit the source, and that malicious copying will be pursued through the law. That is a request and a warning, not a grant. A permissive licence would let anyone reuse, republish, or translate the write-ups with attribution; here there is no stated permission at all, so reposting a chapter or mirroring the images carries a risk the maintainer has already signalled they will act on. Translation is the specific case worth thinking about, because the English edition is maintained by the same project as a separate repository rather than left open.

There is also nothing to pin a version against: the repository has no GitHub releases, and the last push is dated 2026-08-03. If you are a contributor, the path is the join guide, and the note about contributions says the explanations here are primarily C++ with Java, Python, Go and JavaScript implementations added by contributors, which means the multi language coverage is uneven by design and a reader in another language should check the file before planning around it.

Editorial conclusion

Use it as a syllabus if you want one ordered path through array, linked list, hash table, string, stack and queue, binary tree, backtracking, greedy, dynamic programming and graph topics, and budget for repeated problems because technique chapters re-list them. Do not expect code you can execute, a licence you can rely on, or an English edition here, since the top level has no LICENSE file, the reuse rule is a request in the readme, and the English material lives in the keetcoder repository.

Frequently asked questions

Official sources

1. Issues
2. README
3. youngyangyang04/leetcode-master on GitHub