# gold-miner: one Markdown index per category, translator credit per line, last push 2024-04-17

> gold-miner is the Juejin Translation Plan index, a repository of Markdown files listing English technical articles translated into Chinese, with the translator credited on every entry. Nothing in it needs installing, and the format is stable enough to clone, while the last push dated 2024-04-17 makes it an archive rather than a feed.

**xitu/gold-miner** — 🥇掘金翻译计划，可能是世界最大最好的英译中技术社区，最懂读者和译者的翻译平台：

- Repository: https://github.com/xitu/gold-miner
- Website: https://juejin.cn/tag/掘金翻译计划
- Stars: 34,367 · Forks: 4,984
- Language: Unknown
- License: not declared
- Published: 2026-08-17 · Updated: 2026-08-18 · Language: en
- Canonical page: https://hysenlabs.com/projects/xitu-gold-miner

## The repository is an index you read, not a package you install

There is no install step, no build, and no package manifest here, because the deliverable is the text itself. The root is a set of category files: `AI.md`, `algorithm.md`, `android.md`, `backend.md`, `blockchain.md`, `design.md`, `front-end.md`, `integrals.md`, `ios.md`, `others.md`, and `product.md`. Each one is a list of links, and each one closes with a pointer to itself for the full set, in the form of a line reading that all translations in that category continue over. Beside them sit `article/`, `glossary/`, `TODO/`, and `TODO1/`, plus `a-new-post-template.md` and a code of conduct.

The practical consequence is that there is no way to subscribe. Git clone gives you a snapshot of Markdown, and a category file is read in a browser or through a text editor. There is no feed, no index API, and no machine-readable list of what is new, so a team that wants to know what appeared this month has to watch the commit history by hand.

## Credit is a per-line name, and the scoreboard lives in the wiki

Every entry in a category file names the person who translated it, and the name links to a GitHub profile. The convention is uniform: the article title links out to the Chinese version, then a parenthetical gives the translator's name and links to their account, followed by a marker that the entry is a translation. Reading one line tells you the source, the destination, and who did the work, which is why the file works as an index without a separate credits page.

What it does not produce is a contributor list you can query. Ranking lives in a translator points table on the project's wiki, one repository over from the files, so the scoreboard is outside version control and outside the pull request history. The consequence for a maintainer is that the record of who contributed is split across a file and a wiki page, and a rename or a deleted wiki page leaves the Markdown entries as the only surviving attribution.

## The same article is filed under two different categories

One entry appears twice. The article on popular storage approaches for React Native developers, credited to the same translator, is listed in the Android section and again in the iOS section, with an identical link and an identical name. Nothing marks the second occurrence as a cross-reference, and both sections keep their own pointer to their full list.

This is the cost of hand-maintained category files rather than a tagging system. A cross-topic piece gets filed wherever the person doing the filing was looking, and when that happens the two copies drift: an update applied to the Android file leaves the iOS file pointing at the older version, and a reader who arrives from either index sees the same content without knowing it is the same article. If you are mining this repository for a topic, expect to deduplicate by link yourself.

## Older entries sit on one host and newer ones on another

The links are not uniform. The blockchain entries in the root file point at `juejin.im`, while the AI, Android, iOS, front-end, and back-end entries point at `juejin.cn`. Both hosts appear inside a single category file, and the project homepage itself is a tag page on `juejin.cn`.

Nothing in the repository normalizes this. A tool that walks the index has to treat two hosts as one collection, and a reader who finds an older entry may need to try the other domain if a link does not resolve the way they expect. The split is a small thing on its own, and it only becomes a problem when you want to build anything on top of the list, because the assumption that every link lives under one host is the one you have to check before you write any code against it.

## The recommendation queue is an issue form, deduplicated by hand

Submitting an article is the one contribution path the root file points at, and it goes to an issue template. The form asks for two things: the original link, and a short introduction to the article. The template's own guidance tells the submitter to search for the article on Google first and try to make sure it has not already been translated.

That is the entire duplicate check. There is no tooling behind it, no script comparing new submissions against the category files, and no stated reward for a recommendation that gets accepted. The consequence falls on the volunteer doing the checking: dedup quality depends on whether the person who filed the issue remembered to search, and the person who later merges has to catch what was missed. The root file does not describe what happens to a submission after it is filed.

## The rules that govern output live in a wiki, not in the repository

Five tutorial pages carry the process, and all five are wiki pages rather than files in the tree. There is a page on how to take part in translating, one on how to submit a translation and how to handle later updates, one on how to take part in proofreading and on the correct way to do it, one on how to share an article to Juejin, and one set of translation typography rules. The last of those is a style gate, and it lives in the same out-of-tree location as the rest.

Two things follow. A translator working from a clone has everything about what to read and nothing about how the text should look, because the typography rules cannot be reviewed in a pull request against code. And proofreading is a distinct role from translation, so a second person stands between a draft and publication. If you are evaluating whether a translation is trustworthy, the reviewer is named on the line and nothing more.

## Two TODO directories and a template are the whole contribution contract

The layout carries its own history. There are two parallel directories named `TODO/` and `TODO1/`, which reads as two generations of a staging area rather than one agreed place to put work in progress. A State of Vue report for 2017 is linked from inside `TODO/` as a Chinese edition, so the staging directory has been used as a place to publish from. Alongside them, `a-new-post-template.md` is the one contribution artifact that is version controlled, and `glossary/` holds shared terminology.

There is also a sibling repository. The root file links a separate project for the blockchain sub-topic under a different organization repository, so at least one category has been split out into its own index. The consequence for a reader is that the category files are not the whole corpus, and a search for the most recent blockchain translations needs to know that a second repository exists.

## No releases, and the newest push on master is 2024-04-17

The repository is not archived, and it has no GitHub releases at all, so there is no version to pin and no changelog to read. The last push to the master branch is dated 2024-04-17. What the content shows matches that date. The documents section still offers the front-end developer guides for 2017 and 2018, the State of Vue report for 2017, the TensorFlow and ML Kit document sets, and a mirror domain for the Next.js documentation. Nothing in the file has been dated more recently.

The summary line near the top claims more than 4000 translated articles, 13 official documents and manuals, and more than 1500 translators contributing translation and proofreading. Those numbers carry no timestamp, and the documents section below them shows 14 entries, one more than the 13 stated. The honest reading is a well-formed archive with a format that still works and a contents list that stopped moving in 2024.

## Conclusion

Treat gold-miner as an archive to search rather than a source to follow: it needs no tooling, its rules live in a wiki outside the repository, and its newest push is dated 2024-04-17 with no tagged releases to track. Contributors should read the five wiki tutorials before filing an issue, and readers should expect cross-topic articles to appear under two category files and links to sit on two hosts. Before trusting a category file, check that its editions are still current, because the documents list still points at 2017 and 2018 handbooks.

## FAQ

### How much does a gold miner make, according to the gold-miner project?

This repository says nothing about mining or mining income. xitu/gold-miner is the Juejin Translation Plan, a set of Markdown index files that list English technical articles translated into Chinese, with the translator credited on each entry.

### Are gold miners a good investment now, per gold-miner?

There is no market data, pricing, or investment content anywhere in the repository. It is a curated index of translated technical writing, split by topic into files such as AI.md, android.md, and front-end.md.

### how to become a gold miner through the gold-miner project

The path is translation, not mining. The root file links five wiki tutorials covering how to take part, how to submit a translation and later updates, how to proofread, how to share an article to Juejin, and the typography rules, and article recommendations go through an issue template asking for the original link and a short introduction.

### What are gold miners called in the gold-miner repository?

In this repository the name refers to a translation community rather than to mining. The project is the Juejin Translation Plan, its tag page lives on juejin.cn, and the root file links a separate repository holding the blockchain sub-topic index.

## Sources

- [Official documentation](https://juejin.cn/tag/掘金翻译计划)
- [Official README](https://github.com/xitu/gold-miner#readme)
- [Project repository](https://github.com/xitu/gold-miner)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/xitu-gold-miner