# The plan says "week 1 100 stars," and the stars were bought: an AI-built Chan theory workstation · GitHub open-source project

> Source: https://stringers.io/en/github/gh_JustinGongjj__chan-structure-workstation.html (Stringers / 什么值得做, generated 2026-11-20)
> Stringers tracks which small products suddenly took off each day, investigates why one by one, and distils playbooks from them. Numbers come from 15 databases we collect daily, each with source and date.
> For agents: these investigations and the underlying databases (verified revenue, App Store / Google Play charts, ad libraries, GitHub, Product Hunt…) are queryable live over MCP — tools feed (daily breakouts) / plays (playbooks) / hot / products / data. Setup: https://github.com/komako-workshop/stringers (the user signs in with Google at stringers.io/en/start.html to get a personal URL). Site index: https://stringers.io/llms.txt

> It didn't take off: the author had AI build code and a full launch plan over two weeks, never posted anything, and from Nov 6 a batch of bulk accounts gave it 228 stars over several days; similar Chanlun repos barely moved that same week.

[GitHub open-source project](https://stringers.io/en/github/index.md) · written 11-13

JustinGongjj/chan-structure-workstation · JustinGongjj

# The plan says "week 1 100 stars," and the stars were bought: an AI-built Chan theory workstation

**What it is.** A zero-dependency JavaScript Chan theory tool: it automatically computes fractals, strokes, central zones, divergence, and buy/sell points from candlestick charts. It comes with a web trading desk covering A-shares, HK stocks, US stocks, crypto, and futures, plus on-chart teaching annotations. It also packs in divination-based timing and small mini-games.

**Why it went viral.** It didn't. The author used AI for two weeks to build the code and a full launch plan, never did any promotion, and from Nov 6 a batch of bulk accounts gave it 228 stars over several days. Comparable Chan theory libraries in the same week barely moved.

**D** **How you could follow it.** Don't copy it as is. If you really want to build in this space, build only a teaching tool that walks users through Chan theory concepts step by step on their own charts, or wrap a mature engine as an AI-callable skill / MCP, then acquire users on Bilibili, Xueqiu, and Chan theory study groups with screen recordings.

D grade: over 90% of the stars were bought, so they're not a demand signal. In open-source Chan theory there are already older libraries with thousands of stars, like czsc and chan.py, and retail traders use TongDaXin and 10jqka. Copying this exactly won't earn real stars. Few people know about it yet.

Data (updated daily) · Repo created 2026-10-27 · 442 stars · 7 days +238

GitHub · stars

Data 2026-11-07 → 2026-11-20 (14 days, 14 observations). Points before 10-28 come from deeply / archive backfill, not every day has one

chan-structure-workstation is an open-source Chan theory chart workstation. It got 243 stars in 18 days of existence, and 200+ of them came in one week. Looking at the star list, the conclusion is clear: 228 (94%) came from the same batch of bulk-registered accounts with 0 followers. No one outside mentioned it, and it has only 1 forks. This piece covers what it is, where the stars came from, whether there's a real opening in this niche, and what people who want to earn stars for an open-source project can learn from it.

### The first star, Nov 6 morning

On 2026 Nov 6 09:48 (UTC), an account in the format `accrualanxiety8h` starred JustinGongjj/chan-structure-workstation. These accounts were registered between 8 and October, each with 0 followers and 0 following, and each has starred roughly 95 repos. Their names follow the pattern "word + word + a few letters or digits": `00cactuscanoes`, `barrel-stop-4s`, `seaward95twins`.

That day 31 such accounts arrived, the next day 42, and the day after 55. The repo's stars rose from 11-07 to 32 by 11-10 and to 176, then leveled off around 200 two days later. 11-13 then brought 36 more of the same accounts, reaching 238 by 11-14 (GitHub, 11-14). This is the common pattern of star-buying services: place an order, release it slowly over several days so it looks less conspicuous, then place a second order a few days after the first finishes.

Ironically, the repo ships its own launch plan, `docs/LAUNCH_PLAYBOOK.md`, and section one says: "Stars arrive in bursts tied to posts, not continuously."

### What it is

A Chan theory (缠论) tool written in JavaScript with no third-party libraries:

- **Engine** : about 900 lines of code that run any K-line data through Chan theory step by step: inclusion handling, fractals, strokes (笔), segments (线段), pivots (中枢), divergence (背驰), up to 1, 2 and 3 buy/sell points and interval nesting (区间套). Same input, same output, every time.

- **Charts** : a Node service plus a web page that pulls from free data sources (Tencent, Sina, East Money, Nasdaq, Binance, FRED), covering A-shares, HK stocks, US stocks, crypto, China futures and US futures.

- **Teaching** : click a concept (say, 背驰), and the chart highlights the matching segments step by step and shows where the stop-loss line goes. There's also a glossary of 40 concepts and an index of 108 lessons.

- **Extras** : factor backtesting, portfolio performance attribution, "divination-based timing" (scoring with seven methods: Four Pillars, Plum Blossom, Six Lines, Zi Wei, Qi Men Dun Jia, plus a day-by-day backtest), a 108-level mini game, and a mobile PWA with iOS and Android skins.

All of this was built in two weeks, from Oct 27 to Nov 9: 84 commits, most signed `Chan Station <dev@chan-station.local>`, version numbers jumping from 1.0 to 3.12, and some days with a dozen-plus releases. Commit messages say "review agent two BLOCKERs," the Chinese README still contains a sandbox path `/home/node/default/chanlun-station`, and a doc line reads "the runtime has no network for git push, so publish via the GitHub API." The author was directing an AI agent from a cloud sandbox.

The author, JustinGongjj, registered 2024 years ago and has 0 followers. His other repos are small quant libraries created in the same two weeks, each with 0 to 1 stars. He has a demo page on Hugging Face with 0 likes.

### What the plan said, and what happened

The launch plan is very professional. First, a one-line pitch. Then a checklist to get the README, bilingual docs, CI, Docker, citation format and issue templates ready (all of which these repos do have). Then posts go out in 6 waves to Hacker News, Reddit, X, V2EX, Zhihu and awesome lists. The targets are "1 stars by week 100, 4 stars by week 300, 8 stars or more by week 500". Section 6 is titled "Path to 'core contributor of a known framework'", meaning becoming a core contributor to a well-known framework.

The preparation was done, but no launch posts appeared:

- No Show HN on Hacker News;

- Searching X for "Chan theory" and "chan-structure-workstation" over the last 30 days returned 0 posts;

- Nothing found on V2EX, Zhihu, Juejin or Xiaohongshu;

- Web search only returned the GitHub page itself and the Hugging Face demo with 0 likes.

The target numbers weren't hit by posting. They were hit by buying.

### Where the stars came from

I checked the 243 stars account by account:

| Source | Quantity | Pattern |
|---|---|---|
| Batch accounts | 228 | Registered from 2026-08 to October (63 in August, 73 in September, 98 in October), 0 followers each, each starred 64 to 110 repos, median 95 |
| Acquaintances on the day the repo was created | About 6 | 10-27 showed up that day. One account was registered that morning and starred only this repo. The rest are small accounts with 0 to 2 followers. |
| A few older accounts and the author himself | About 9 | The author starred his own repo 11-09 time(s) |

The only fork came from a small account of 11-09, whose name includes "Liverpool," the same as one of the accounts that cheered on the creation day. My inference is that it's the same acquaintance. 0 issue(s), 0 watcher(s).

Now look at the same week's peers. czsc (6,600 stars), the biggest open-source Chan theory library, grew by 28 that week, chan.py grew by 8, chanlun-pro by 8, and the other Chan theory libraries mostly sat between 0 and 3. The topic didn't heat up. Only this one grew.

These batch accounts weren't set up specifically to inflate this repo. I sampled 40 of them and looked at what they had starred recently: 679 distinct repos in total, mostly Chinese AI tool repos created in 9 or October, with star counts pushed to around 500 to 1,500. A run of repos starting with `mcd-`, created the same week, were starred by nearly all of these 40 accounts. My guess is some star-count contest, but I haven't verified it.

### The Chan theory slot: people use it, but the positions are taken

Setting the star-buying aside, Chan theory is a real topic. The rules are complex and hard to learn, a large group of retail traders are studying it, and open-source implementations have had users for years. But most of the obvious positions are already filled:

| Project | Where | Launched | Scale | How it differs |
|---|---|---|---|---|
| waditu/czsc | Open source | 2019-06 | 6,596 stars / 1,811 forks | The de facto standard in the Python quant community, with no web or mobile version |
| Vespa314/chan.py | Open source | 2021-12 | 2,312 stars / 845 forks | The most complete Python framework, with interval nesting and an ML interface |
| yijixiuxin/chanlun-pro | Open source | 2021-12 | 1,094 stars / 394 forks | Full-featured, heavy to deploy locally, partly paid |
| ibaihuo/chanvis | Open source | 2022-04 | 630 stars | Focused on TradingView-style visualization only |
| ChanlunX, rust-chan and other TDX plugins | Open source | 2016–2023 | 350–430 stars each | Installed directly into TDX (通达信), the mainstream setup retail traders actually use |
| Chan theory AI Skills (czsc_skills and similar) | Open source | 2026 | 20–145 stars | Packaged as agent skills so an AI can explain Chan theory |
| Chan theory indicators for Tonghuashun, TDX and TradingView | Charting software | Years | Retail mainstream | Built in, zero setup |
| Chan theory drawing | iOS | 2026-05 | China store 3 ratings | The only standalone Chan theory app I could find, with essentially no traction |

(GitHub, App Store, 11-14)

Here's a number that separates real from fake: most Chan theory libraries created this year have a dozen to a few dozen stars, but forks are 20% to 60% of stars, because people who actually use them fork them to tweak parameters. This repo has 243 stars and 1 forks, which is 0.4%.

### What to learn from it

**A good launch plan, and a cautionary result.** The LAUNCH_PLAYBOOK is worth reading: a one-line positioning, different hooks for different audiences, a checklist before posting, posting in 6 waves, and a weekly list of numbers to watch. Its only flaw is that it wasn't executed. AI can write the code, docs, CI and competitor tables in a week, but distribution is still something a human has to do.

**When you look at a new repo, don't look at stars.** Check the fork-to-star ratio, whether real people have filed issues, whether anyone is watching it, and whether someone outside the project has written about it. If you can see the stargazer list, check when those accounts were registered and how many followers they have. That will usually show you right away whether they're batch accounts.

**A good idea nobody has validated.** "Click a concept and have it marked step by step on the chart you're looking at" is a real pain point for a hard method like Chan theory. Open-source Chan libraries compute structure, but few teach it. Still, no real person has used this one, so the size of the demand is unknown.

### If it were you

**Verdict: D. Take a look, but don't copy it.** Its 243 stars are not a demand signal; they were bought. Its real user base is probably in the range of a dozen people (my inference). Building another "Chan engine plus multi-market charts" won't earn real stars. In this space you have older libraries like czsc and chan.py with thousands of stars, and on the retail side TDX and Tonghuashun.

**Who it suits** : developers who already understand Chan theory, already teach it on Bilibili, Xueqiu, Zhihu or WeChat Official Accounts, and have a group of students or readers.

**Who it doesn't suit** : people who don't understand Chan theory, people who want to pile up features with AI and wait for stars, and people planning to start by buying stars. Bought stars won't turn into forks, issues or shares, and they won't make people see you as the go-to answer for "how to code Chan theory."

**If you still want to build it, go after the piece it didn't do well:**

1. **Teaching annotations only, done thoroughly.** Skip divination and games. Just do "give me a K-line chart and tell me step by step where the fractals are, where the strokes are, and why divergence happens here." Build on czsc or chan.py underneath. That saves you writing an engine and saves users from doubting whether your algorithm is right.

2. **Or make it a tool an AI can call directly.** Wrap a mature Chan theory engine as a skill or MCP that Claude Code or Codex can use, so someone can ask in a chat, "which buy/sell point is this stock at now?" This slot has only a few dozen to a hundred-plus stars today and no clear leader yet, but the ceiling is also not high.

**How to get users** : people interested in Chan theory are on Bilibili, Xueqiu, Zhihu and study groups, not on Hacker News. Post a screen recording that explains divergence in one chart, and answer specific questions in creators' comments and in groups. Once you have a batch of real users, let forks and issues grow on their own.

**Timing** : there's no window here. The topic won't expire, and it won't suddenly heat up either.

**What would show this read is wrong** : (1) real people start writing about this library, filing issues or PRs, and forks climb into the dozens, which would mean real use and I've underestimated it; (2) someone ships an open-source tool focused only on teaching Chan theory annotations, gets to 1,000 stars in three months without buying any, which would mean teaching is worth more than I've estimated here and the grade should go up.

### Data and sources

GitHub: JustinGongjj/chan-structure-workstation repo page, commit history, release history, `docs/LAUNCH_PLAYBOOK.md`, `docs/COMPETITIVE.md`, `docs/PUBLISH.md`, `README.zh-CN.md`. Stargazer accounts come from the repo's public event records (245 starring events, with accounts and timestamps). Account registration dates, follower counts and number of repos starred come from public GitHub profiles (pulled 11-14). Daily star counts are logs we recorded each day (11-07 to 11-14). 11-13 current figures: 243 stars, 1 forks, 0 watchers. Peer repo figures were checked live on 11-14. Hugging Face demo: 0 likes (11-14). Hacker News, X and web search results are as of 11-14. I did not find any star sellers or their prices. Whether the `mcd-` repo series came from a contest is unverified. The author's motive for buying stars (job hunting, résumé credibility) is only speculation.

[GitHub repo ↗](https://github.com/JustinGongjj/chan-structure-workstation)

**Research notes: competitor table, acquisition evidence, full research report**

Competitors

| Name | Where | Launched | Scale | How it differs from this one |
|---|---|---|---|---|
| waditu/czsc | Open source | 2019-06 | 6,596 stars / 1,811 forks | Long-established Python/Rust quant libraries, the de facto standard; no web or mobile product and no teaching annotations |
| Vespa314/chan.py | Open source | 2021-12 | 2,312 stars / 845 forks | The most complete Python framework (interval nesting, ML interface); no frontend product |
| yijixiuxin/chanlun-pro | Open source | 2021-12 | 1,094 stars / 394 forks | Full-featured local tool, heavy to deploy |
| ChanlunX / rust-chan / neuks/Indicator (TDX plugins) | Open source | 2016-2023 | each 350-430-star | Drop it straight into TongHuaShun, the mainstream retail-trader setup. It's also a standalone web app. |
| Chan theory AI Skill repos (noahnan-max/chanlun-trading-system, zengbin93/czsc_skills, etc.) | Open source | 2026 | 20-145 stars | Ships as an agent skill so an LLM can explain Chan theory. No skill/MCP here. |
| Chan theory indicators for TongHuaShun, TongDaXin, and TradingView | Big-company features | Years | Retail mainstream | Built-in charting software, zero setup |
| Chan theory drawing | iOS | 2026-05 | cn store 3 ratings | The only standalone Chan theory app, barely any traction |

Where users come from

- **Verified** **Bought GitHub stars (star farm)** : the events endpoint returned 245 WatchEvents. 228 accounts were created after 2026-08, have 0 followers and 0 following, star-to-repo median 95, and use generated usernames (accrualanxiety8h etc.). They started starring in batches from 11-06 09:48 UTC, matching the star-growth curve. Sampled 40 accounts: together they starred dozens of new Chinese repos, including mcd-party-planner, Deepscient, and axuse.

- **Verified** **Friends showing support** : 10-27 stars on the day the repo was created, 6 of them from gyc-1016, created that same day and starring only this repo. The rest are mostly 0-2 follower alt accounts.

- Inferred **HN/Reddit/X/V2EX/Zhihu posts** : the repo includes a LAUNCH_PLAYBOOK.md with a full plan, but we found no launch posts on HN, x-radar, Jina, Brave, or Exa.

## JustinGongjj/chan-structure-workstation: why it gained stars

### Bottom line first

**This repo was pumped with purchased stars, not organic buzz.** Of 243 stars, 228 (94%) came from one batch of bulk-created alt accounts: 2026-08 created in October, 0 followers and 0 following each, each having starred roughly 95 repos, with generated usernames like `accrualanxiety8h` and `acerbic-case-3p`. These accounts began starring in a burst at 2026-11-06 09:48 UTC, matching the two growth periods we saw, 11-07 to 11-10 and 11-13. About 15 more stars came from friends on the day the repo was created (10-27) and in the first few days, many of them brand-new accounts that starred only this repo. We found no posts, reposts, or press coverage off-platform. The HF Space has 0 likes. The repo has only 1 forks, 0 issues, and 0 watchers, and the author has 0 followers.

The product is a kitchen-sink project one person built by directing AI agents: a Chan theory engine plus multi-market charts, educational annotations, fortune-telling timing, a quiz game, and a dual-skin PWA. The repo's `docs/LAUNCH_PLAYBOOK.md` lays out a full HN / Reddit / X / V2EX / Zhihu star campaign with targets of 1 stars in week 100 and 4 stars in week 300. We found no evidence any of these channels were actually executed. The star count was bought to hit those targets.

What it means for us: this isn't a product signal worth following. It's a case study in a **star-farming supply chain** : the same batch of accounts is starring dozens of new Chinese repos at once. See section 4.

---

### 1. What it is

- Repo: [github.com/JustinGongjj/chan-structure-workstation](https://github.com/JustinGongjj/chan-structure-workstation). Created 2026-10-27, JavaScript, self-declared MIT (GitHub detects NOASSERTION), latest release v3.12.1 (2026-11-09).

- Contents: about 900 lines of zero-dependency JS for a Chan theory structure engine (processing → fractals → bi → segments → pivots → divergence → 1/2/3 buy/sell signal types, nested intervals), wrapped in a Node server and a web workstation. It covers six markets (A-shares, HK, US stocks, crypto, China futures, US futures) using free data sources (Tencent, Sina, EastMoney, Nasdaq, Binance, FRED). Other features:
    - "Concept → on-chart annotation" teaching: click a concept and the chart highlights a+A+b, the two divergence legs, and the stop-loss line step by step;
    - 40 concept glossary entries, 108 lesson index, and a "quotes collection";
    - Factor backtesting and quant portfolio performance attribution;
    - "Fortune-telling timing": Four Pillars, Plum Blossom Numerology, Six-Line Najia, Zi Wei daily flow, and Qimen Dunjia, scored in parallel with seven methods and backtested day by day;
    - 108 levels of a small quiz game;
    - Mobile PWA with separate iOS and Android skins.

- How it was built: 84 commits, most under the author name `Chan Station <dev@chan-station.local>`. Commit messages include "review agent two BLOCKERs"; `docs/COMPETITIVE.md` has the line "benchmarked against the Zi Wei Dou Shu you mentioned"; `docs/PUBLISH.md` says "the runtime has no network for git push, so publish via the GitHub API"; and the Chinese README still contains the sandbox path `cd /home/node/default/chanlun-station`. It looks like the author directed AI agents in a cloud sandbox, shipping roughly a dozen versions a day from 10-31 to 11-09, with version numbers climbing from 1.0 to 3.12.

- Author: JustinGongjj, account registered 2024-06, 0 followers. The other 5 repos are a "Merrill Lynch clock" quant repo and an empty repo, both created between 10-31 and 11-10, each with 0 to 1 stars. Hugging Face account JustinGong2026 has two static Spaces, each with 0 likes (HF API, 2026-11-14).

### 2. Growth data

Our GitHub repo (repo_day, observed from 11-07):

| Date | stars | Daily adds |
|---|---|---|
| 11-07 | 32 | — |
| 11-08 | 80 | +48 |
| 11-09 | 139 | +59 |
| 11-10 | 176 | +37 |
| 11-11 | 197 | +21 |
| 11-12 | 199 | +2 |
| 11-13 | 204 | +5 |
| 11-14 | 238 | +34 |

- Similar repos did not rise together: the same week, czsc (6.6k stars) gained only 28, chan.py gained 8, chanlun-pro gained 8, and the other Chan theory repos were mostly between 0 and 3 (GitHub repo `search '缠论|chanlun|...' --sort gain`, 11-14). So this isn't a topic heating up.

- It never appeared on trending (it was under 60 days old, so it couldn't have). Our repo's suspect column is empty: this repo didn't trigger the rule "≥300 stars and forks≤1" because it has only 243 stars.

### 3. Where the stars came from: we checked each starring account

GitHub's stargazers endpoint returned 404 for our token, so we used the repo events endpoint instead. It contains 245 WatchEvents total, with the starrer and timestamp. We then pulled profile data for these 244 accounts via GraphQL (1 are deleted).

- **228 are farm accounts** , defined as: created after 2026-08, 0 followers, and starred more than 20 repos. Across these 244 accounts, 238 have 0 followers and 238 have 0 following. The median number of repos starred is 95 (p10 to p90 is 64 to 110). The median number of repos they own is 4. Account creation clusters in 2026-08 (63 accounts), 09 (73 accounts), and 10 (98 accounts).

- Usernames: 234 are generated-style, such as `00cactuscanoes`, `accrualanxiety8h`, `barrel-stop-4s`, and `seaward95twins`.

- Star timing: the farm accounts' first star landed at 11-06 09:48 UTC. After that, stars spread across days: 11-06 31, 11-07 42, 11-08 55, 11-09 37, 11-10 18, 11-11 3, 11-12 4, 11-13 36, and 11-14 2. This is a classic "order, then drip over several days" pattern, with the second order arriving on 11-13.

- About 15 stars from accounts other than the farm: 10-27 got 6 on the day the repo was created. Among them, `gyc-1016` signed up that same day 08:54, while `yrui430`, `jinouy`, `Finnfuture` and `anfield59457-collab` are small accounts with 0 to 2 followers, which looks like friends cheering it on. After that come a few scattered old accounts, plus the author's own stars (11-09). The only fork is from `liverpooler001` (11-09), and like `anfield59457-collab` it carries a Liverpool name, so it may be the same circle.

- We searched off-platform and found nothing:
    - Jina, Brave, and Exa searches for the repo name and Chinese name returned only the GitHub page itself and the HF Space;
    - x-radar searches for "缠论" and "chan-structure-workstation" returned 30 posts in the last 0 days;
    - Our HN repo has no matching Show HN;
    - Searches on V2EX, Zhihu, Juejin, and Xiaohongshu found no trace of the repo;
    - The awesome-list PRs and other actions listed in LAUNCH_PLAYBOOK also left no trace.

**So the only growth tactic we can confirm is buying GitHub stars.** The launch plan looks like an AI-written template with no evidence it was executed.

### 4. Who else this farm is starring (found along the way)

We randomly sampled 40 farm accounts and pulled the repos they've starred recently, to see which ones got hit repeatedly. Most-starred (in brackets: how many of the 40 accounts starred it, and the repo's star count at the time):

- zcbacxc/mcd-party-planner (all 40 accounts starred it, 1382 stars)

- yzlin286-wq/Deepscient (36,984 stars)

- ctoldigital/axuse (36,804 stars)

- chuiziyao/maimen-certification (35,1028 stars)

- wuyoscar/Internal-Safety-Collapse (35,2619 stars)

- bugkeep/ai-draw (34,985 stars)

- flybrainlabs/miner (34,968 stars)

- Alice-Marx/dsh-model-router (34,962 stars)

- TonF1917/claude-desktop-multiple-windows-different-api (33,984 stars)

- Rice-dog/code-codex (33,1578 stars)

- hujjun/mcd-saver-strategist (33,1333 stars)

- shx2005/codex-account-switcher (32,970 stars)

- Sam-Hui-dot/mcd-budget-meal (31,1327 stars)

- yeqv111/mccarbon (28,1793 stars)

- lvzuojing/mcd-wrapped, spideymooon/McOracle, AAA-EE/mcd-budget-planner, and others

The 40 accounts together starred 679 different repos. The pattern is clear: the customer is buying stars in bulk for new Chinese AI and tooling repos created from 2026-09 to October, pushing them to around 500 to 1500 stars. A cluster of `mcd-*` repos (created around Oct 9 to 15) suggests a star-ranked contest, possibly a McDonald's-themed developer challenge, whose entrants bought stars collectively. This is unverified. Our github repo's suspect column is empty for all of these repos, so current rules don't catch this farm. If we want to clean recall data, co-starring analysis across these accounts is a workable way to identify it.

### 5. Competitors

| Name | Where | Launched | Scale | How it differs |
|---|---|---|---|---|
| waditu/czsc | Open source | 2019-06 | 6,596 stars / 1,811 forks | An established Python/Rust library tied to tushare, the de facto standard in the quant community. This one adds a web UI, mobile app, and educational annotations. |
| Vespa314/chan.py | Open source | 2021-12 | 2,312 stars / 845 forks | The most complete structure in a Python framework, with nested intervals and ML interfaces. No frontend product. |
| yijixiuxin/chanlun-pro | Open source | 2021-12 | 1,094 stars / 394 forks | Feature-rich, leans toward local deployment, heavy repo; some features paid. |
| ibaihuo/chanvis | Open source | 2022-04 | 630 stars | just a TradingView visualization shell |
| ChanlunX / rust-chan / neuks/Indicator | open source (Tongdaxin plugin) | 2016–2023 | 350–430 stars | installs straight into Tongdaxin, the mainstream form retail traders actually use |
| noahnan-max/chanlun-trading-system, adsorgcn/chanlun-engine-skill, zengbin93/czsc_skills | Open source (AI Skill) | 2026 | 20–145 stars | packaged as a Claude/agent skill so AI can explain Chanlun; no skill/MCP shipped (only written into the roadmap) |
| chanlunym666/chanlun-engine | Open source | 2026-08 | 67 stars / 38 forks | a Rust DLL for Tongdaxin/MT4; the data card calls it a 'same-product reskin', but its description and implementation both differ, so it's a misreading |
| TensorCode666/stock-chanlun | Open source | 2026-04 | 85 stars / 45 forks | a Vue+FastAPI+LLM A-share analysis system, also not the same product |
| Chanlun indicators for THS / Tongdaxin / TradingView | big-company features, community scripts | Years | Retail mainstream | built into the charting software, zero setup |
| Chanlun drawing (iOS) | iOS | 2026-05 | cn store 3 ratings | the only standalone Chanlun app you can find, barely any usage (appstore library) |

Since AI started writing code 2026, a lot of Chanlun repos have appeared. Most have a few dozen stars at most, with forks at 20% to 60% of stars. People who actually use them fork them and modify. This repo has 243 stars and only 1 forks, which is extremely abnormal for its category.

### 6. Differentiation

Its self-described differentiators are written in `docs/COMPETITIVE.md`: zero dependencies, determinism, free multi-market data, concept→on-chart annotation, a mobile PWA, and backtestable timing. From the feature list, that is indeed rare among open-source Chanlun projects: the engine, product, and teaching material live in one repo.

**But these differentiators are not why it gained stars.** The star growth has no relation to any real spread: no posts, no forks, no issues, and the category itself isn't heating up. The feature sprawl, with fortune-telling, Zi Wei, Qimen, and a game mixed together, actually dilutes the 'rigorous and verifiable' positioning. Its real audience is on the order of a dozen people.

### 7. What to borrow

- A cautionary example: an AI agent can generate in a week a repo with extremely broad features and complete docs, including CITATION, CI, a release playbook, and a competitor table, yet distribution stays at zero. Stars can be bought, but you can't buy forks, issues, or off-platform discussion. When you look at a new repo, fork/star ratio, watchers, and the registration age and follower count of starring accounts are far more reliable than the star count itself.

- Method: use the events API to pull starrers, then profile accounts and run co-occurrence clustering. That can identify a whole star-farm operation and its client list in one pass. It could be added to the suspect rules in our github library, e.g. 'median account age of starrers under 90 days and 0 followers' or 'overlap with known farm accounts'.

- The only product idea worth noting: teaching-style annotation, 'concept → marked step by step on your chart', is a real need for a hard methodology like Chanlun. But nothing shows this repo has anyone using it.

### 8. Uncertain

- I couldn't find the specific star sellers or their prices (presumably GitHub star-boosting services sold on Taobao, Xianyu, and similar sites).

- Whether the series of `mcd-*` repos is tied to a contest that awards prizes by star count is unverified.

- The starrer list comes from the events API. That API only goes back 300 events and 90 days. All 256 entries here fall within that window, but a few canceled stars may not be counted.

- The author's motive for buying stars (job-hunting credibility? padding a 'core contributor' résumé? LAUNCH_PLAYBOOK section 6 describes exactly this route) is speculation.

### Sources

- GitHub REST: `repos/JustinGongjj/chan-structure-workstation`, `/events`, `/commits`, `/releases`; `docs/LAUNCH_PLAYBOOK.md`, `docs/COMPETITIVE.md`, `docs/PUBLISH.md`, `README.zh-CN.md`; GraphQL account profiles (2026-11-14 pulled).

- Our github library: repo_profile, repo_day (2026-11-07 through 11-14).

- HF API: `api/spaces/JustinGong2026/chan-structure-workstation` (likes 0, created 2026-11-01).

- Jina, Brave, Exa search, x-radar, hackernews library, appstore library (2026-11-14).

This article was written by AI from public data and the data we collect daily. Numbers are sourced, and inferences are labeled.
