Stringers
ksjhk/api · ksjhk

A 26KB API doc that gained 160 stars in a week: Kashijie used GitHub as the storefront for recruiting agents

What it is The open platform API docs and PHP/Python/Node/curl examples for Kashijie (kashijie.com), a card voucher site. Plug them in and you can check membership card floor prices, place recharge orders, and auto-deliver.

Why it took off Developers didn't bookmark it on their own: the kaquan repo the author promoted on Juejin has only 1 stars, while this unpromoted doc gained +160 in a week, with 0 forks, 0 watchers, and 0 issues. Meanwhile a batch of Chinese repos doing the same business look nearly identical. The stars serve to rank it first for searches like "card voucher supply API" and make it look like a legitimate open-source project.

DWhat you can do Don't copy this by writing a doc or faking stars; you won't get real stars, and it damages your reputation. What's worth borrowing is one search term, one example repo, and a README written so a coding agent can run through three or four steps successfully.

Grade D: the stars are a facade bought through a gift-card business, not something the project grew on its own. Developers who try to copy this to build stars won't get real stars, since star-farming violates GitHub's terms. Only two ideas are worth knowing: "GitHub as a B2B landing page" and "README written for agents." Almost nobody knows about it yet
Data (updated daily) · Repo created 2026-10-26 · 373 stars · 7 days +225
GitHub · stars
020040011-0711-1011-1311-1711-20373
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

On GitHub there's a repo called ksjhk/api containing just one API doc for a card voucher supplier and call examples in four languages. It gained 193 stars in 19 days, 160 of them in the past week. But it has no forks, no issues, and not even the author watches it. Behind it is a supply platform recruiting "card voucher promotion agents", so the stars were most likely not given by developers. This piece explains what it is, where the stars came from, why a company selling membership cards would farm stars on GitHub, and what someone wanting to grow stars for their own open-source project can take away from it.

A Juejin article from Nov 9

On the evening of Nov 9, 2026, a post appeared on Juejin titled "Can't get supply for membership cards and vouchers? I open-sourced a distribution platform: 200+ brand APIs, auto-delivery." It opens in the familiar indie-dev voice: "Started tinkering with a side hustle this year: sourcing membership cards, hotel bookings and movie tickets, and building my own distribution platform... There are plenty of supply platforms out there, but they're either closed to individuals or their API docs are minimal at best." The post ends with a GitHub link ksjhk/kaquan and a line: "If you find it useful, drop a Star."

The kaquan repo this post pushes is a deployable Laravel backend plus static frontend, 1.8MB of code. As of Nov 13, it has 1 stars.

The same author's other repo, ksjhk/api, isn't mentioned in the post. It's just a README plus examples, 26KB in total. The same week it went from 35 to 183 stars (GitHub daily star count, 11-09 to 11-14), while kaquan, the one that was pushed, showed 11-13 stars when the page was opened 193 times.

The pushed repo didn't move; the unpushed one gained 160 in a week. If readers were the ones starring, the pattern wouldn't look like this.

What it is

In one line: the open platform API docs for kashijie.com (卡世界卡券网). Once connected, your program can query floor prices for membership vouchers like QQ Membership, iQiyi, Ele.me red packets and JD Plus, top up the platform balance first, then call the API to place orders, and the platform delivers automatically. It also covers hotel booking and movie-ticket seat selection.

The README highlights three selling points: read-only endpoints you can call without registering, so you can check supply and compare prices first; submitting a phone number auto-creates an account and issues a key; and each key can set a "markup multiplier" from 1.00 to 99.99, so resellers can apply a uniform markup when they resell.

The README says "200+ brand supply". As of Nov 13, calling its public supply endpoint returned 123 products in total. The price gaps are visible: a QQ Membership monthly card with a face value of ¥14 has a floor price of ¥9.93; a Jimeng premium annual card with a face value of ¥10,398 has a floor price of ¥4,962.75 (public endpoint, 11-13).

Looking one layer further out, the kashijie.com homepage says "Direct card-voucher supply, high commissions with instant settlement, easy downline expansion, agent referral riches", plus "10000+ agent student community" and "Live mentor coaching to your first sale" (official site, 11-13; headcount unverified). Product images are served from the CDN of "卡世界号卡" (Kashijie SIM cards), an older platform that distributes phone-number SIM cards, with many agents running their own sites. The GitHub account is labeled "卡世界号卡官方" (official), but there's also an agent site with the same name, ksjhk.online. Whether the account belongs to the company or to a large agent is unverified.

Put plainly, this is a business that recruits side-hustle agents and has them top up first, and the GitHub repo is one of its entry points.

Where the stars came from

DateCumulative starsDaily gain
11-0723—
11-0827+4
11-0935+8
11-1053+18
11-1177+24
11-12124+47
11-13148+24
11-14183+35

Source: GitHub daily star counts. Forks stayed at 0 throughout.

GitHub no longer publishes the list of accounts that starred a repo, so it's impossible to check who starred it one by one. Any conclusion is an inference, based on these clues:

  • Same author, 11 repos, and only this one has stars. The kaquan repo has more content and actually runs, with 1 stars; 11-09 the newly created hotel-reseller and movie-ticket-reseller are the same doc set with a different keyword, and each has 0 stars.
  • 0 fork、0 watcher、0 issue,作者 1 个 follower,账号 5 月才注册。 只看 0 fork 说明不了什么,近 30 天新建、150 到 250 星的仓库里约三成多也是 0 fork。但一个接口文档仓库 193 颗星,没有一个人提问题、作者自己都没 watch,这在真开发者项目里很少见。
  • No spread found. A search of X over the past 60 days for "卡世界" and "kashijie" returned nothing; there was no Hacker News post, no media coverage, no trending placement. The only traces are its own Juejin post and a kashijie.com snippet surfacing in the search summary of a Zhihu question.
  • A whole batch of similar-looking Chinese repos appeared at the same time. In the last 45 days, repos created recently, with hundreds of stars, 0 forks, Chinese descriptions, and a commercial purpose: those topping the weekly star-gain list include one virtual credit card brand that opened a dozen accounts, each repo with about a hundred stars and roughly +40 a week; and a ChatGPT top-up tutorial repo with 182 stars and 0 forks. ksjhk/api ranks third in this batch.

Another claim is that it piggybacked on the DeepSeek Harness plugin directory. Checking shows that's not true: the repo has no topics, the README never mentions it, and a search of the plugin directory index returns no Kashijie entry (11-13).

Why would a company selling membership cards need stars?

Because stars buy search ranking and first-glance trust.

On Nov 13, searching Brave for "卡券货源 API github" and "卡券 开放平台 API 签名下单 自动发货" put github.com/ksjhk/api as the first result. Searching "会员卡券 API 调用示例 PHP Python" put its own Juejin post first and this repo second. github.com already carries high domain authority, and with a couple hundred stars on top, someone looking to source card supply sees "a legitimate open-source project" rather than an agent-recruiting page.

The card-voucher supply market is crowded. Listed companies like Fulu need corporate documents and signed agreements to join; dozens of mid-sized supply terminals such as Lechen Shuka, Suquanyi and Chengquan all write "free API integration" and push sponsored articles on Baidu and Zhihu. None of them has a proper repo on GitHub. Kashijie opened another channel, and with one repo per keyword (one for hotels, one for movie tickets), the cost is nearly zero.

There's a newer angle too: the first step in the README quick start says "have the user provide a phone number that isn't registered yet, and the program calls /platform/quick-start to auto-register and generate the first set of keys." That sentence isn't written for people, it's written for coding agents like Cursor and Claude. Someone tells an agent "help me connect a card-voucher supplier," and the agent can follow these steps to open an account, get keys, and place orders on its own. The author also built a skill for Kimi and a plugin package for DeepSeek Harness, betting that people will eventually ask agents to hook up supply for them. How many accounts this brought in is not visible from outside.

Is anyone copying it

No one is copying it. In the same niche, several GitHub landing pages appeared at roughly the same time (kamihub, kaquan-nav, rights-price-hub), each with 0 to 2 stars, because they weren't boosted. Zooming out, "Chinese commercial projects using star-boosting landing pages" has become a trend, and this is one of them, not the first.

Open-source card-issuing systems do have real substance: Dujiaoka has 12k stars, and Acg-faka has 5,900 stars. They only do storefront software without supply. A control case is PriceAI, launched in May this year, which compares prices for AI subscription card shops: 5,457 stars and 248 forks, with a normal fork rate, looking like something that grew on its own.

What's still unclear

  • Whether the stars were bought: the list isn't visible, so only indirect evidence exists.
  • Whether the account is Kashijie's official one or a big agent's: not confirmed.
  • How many accounts and top-ups this approach brought in: kashijie.com isn't in the global top 1M sites, and signup numbers aren't visible.

If it were you

D. Worth understanding this playbook, not recommended for copying. Its stars aren't a good project growing on its own; they're a storefront a card-voucher business paid to prop up. Copying the API docs won't earn a single real star; copying the star-boosting puts your own GitHub account at risk.

Who it fits: people running API-based, supply-based, B2B businesses who want one more entry point that developers can find. For you, one repo per keyword, multi-language examples, and a README written for agents all cost nothing.

Who it doesn't fit: individual developers trying to collect stars and reputation through open source. This repo has nothing at the code level worth "rewriting in another language," and copying its 193 stars won't get you anything either.

If you want to borrow something, here's how:

  1. Borrow the "one search term, one repo" idea, but on a real project: if your library has several clearly different use cases, give each one its own example repo, named with words people actually search for.
  2. Borrow the README-for-agents approach: compress "install → get a key → run the first request" into three or four steps a machine can follow, with no captchas and no manual review. More and more coding agents pick libraries for people, and projects an agent can get running in one pass gain a route in.
  3. For a Chinese-language launch, a Juejin post that tells the story in your own voice ("why I built it, how to get it running") plus a repo that runs out of the box is enough to start. The post's writing is fine on its own; the kaquan repo just didn't get starred, which shows an article alone isn't enough. The repo has to be something people want to use.
  4. Don't buy stars. 0 forks, 0 watchers and 0 issues next to a couple hundred stars are obvious to anyone who knows what to look for; GitHub's terms explicitly prohibit it, and star-purge bans happen every year.

How to get users: It goes after search-engine long-tail keywords + Juejin posts + Zhihu mentions. For personal open-source projects, what matters is reaching real developers: Juejin, V2EX, Show HN on Hacker News, and a demo that runs in 30 seconds beat star counts.

When to act before the window closes: no rush, this isn't a window that will close.

What would show this read was wrong: ① Over the next month or two, external issues or PRs appear, or someone publicly says they connected to its API, showing the docs really are in use by developers; ② when searching "card-voucher API" or "supply API", the first page starts filling with similar GitHub repos, showing "GitHub as a B2B landing page" has been proven and is worth copying seriously for people in API businesses; ③ conversely, if GitHub purges most of its stars or the account gets banned, that confirms the stars were manufactured.

---

Data and sources: GitHub repo pages, daily star counts, the same author's other repos (as of 11-14); kashijie.com official site and public supply endpoint (11-13); the Juejin original post 2026-11-09; Brave search results (11-13); X search over the past 60 days. Items marked "inference" or "unverified" are estimates and guesses.

GitHub repo ↗

Research notes: competitor table, acquisition evidence, full research report
Competitors
NameWhereLaunchedScaleHow it differs from this one
Fulu Open Platform / Fulu JujiWebEarly days (Fulu Holdings)Claims to connect 1500+ brands, about 22k productsRequires corporate documents, signed agreements, review; Kashijie lets you browse supply without registering, and submitting a phone number opens an account
Lekayi lekayi.comWebNot verifiedClaims 3000+ SKUs, B2B, OpenAPITargets enterprise procurement; Kashijie targets individual side-hustle agents
Lechen Shuka / XC Quanyi Shuka / Suquanyi / Laohuguo / Chengquan and other supply terminalsWebNot verifiedDozens of mid-sized supply terminals, all writing "free API integration"Also relies on SEO sponsored posts, but has no proper repo on GitHub; Kashijie has claimed the GitHub developer entry point
assimon/dujiaoka (Dujiaoka card-issuing)Open sourceAbout 201812,125 starJust a card-issuing storefront system, no supply; Kashijie's kaquan bundles the storefront with its own supply
lizhipay/acg-faka (Acg-faka card-issuing)Open sourceAbout 2021-105,911 starSame as above: system only, no product sales
dimthink/PriceAIOpen source2026-05-085,457 stars, 248 forksAdjacent niche (AI subscription card-shop price comparison), normal fork rate, a control case for organic growth
Kashijie SIM cards (ksjhaoka)WebAbout 2022A traffic-card distribution platform; many agents run their own sitesSame brand ecosystem for SIM cards; card vouchers are a newly added category
Where users come from
  • Inference GitHub stars prop up credibility (suspected bought or manufactured): 0 forks, 0 watchers, the author has 1 followers; the author's other 10 repos (including the same-content hotel and movie versions) almost all sit at 0 stars; the kaquan repo pushed on Juejin has only 1 stars, while the unpushed api repo gained +160 in the same period; a batch of similar Chinese commercial repos (the CoreGold matrix, etc.) also shows 0 forks and flat curves
  • Verified search engine SEO (borrowing github.com's domain authority): searching Brave for "card voucher supply API github" and "voucher open platform API signed order auto-delivery", the #1 result both times is github.com/ksjhk/api (2026-11-13)
  • Confirmed Tech-community sponsored post: Juejin 2026-11-09 "Can't get card-voucher supply? I open-sourced a distribution platform: 200+ brand APIs, auto-delivery", ending with a request for stars
  • Confirmed Q&A placement: the search summary for the Zhihu question "Are there any reliable virtual card-voucher supply platforms?" shows www.kashijie.com
  • Confirmed Integration docs and skills written for AI agents (GEO): the README has the program take a user's phone number, call /platform/quick-start to auto-open an account and issue keys; also built KSJAPI, ksjskill, copyksjvip (a skill for Kimi), and the DSH plugin in kaquan
  • Verified: agent-network and multi-level distribution: kashijie.com promises "easily expand your downline, earn while you recruit agents" and "10000+ agent students in community"
  • Verified: keyword-matrix of repos: 11-09 built hotel-reseller and movie-ticket-reseller at the same time, same doc templates aimed at different search terms

Why ksjhk/api (the Kashijie open platform API docs) got traction

Research date 2026-11-13 (data card refreshed to 11-14). Sources: our github database (repo_profile / repo_day), the GitHub API, kashijie.com's public endpoints and website, Jina / Brave search and page text, Juejin articles, awesome-dsh-plugin.com's plugins.json / index.json, x-radar.

Bottom line

It's not an open-source project that developers starred on their own. It's a card-voucher supply platform's (kashijie.com, 卡世界卡券网) acquisition landing page hosted on GitHub: a 26KB README plus PHP / Python / Node / curl examples. The 193 stars were most likely bought, farmed, or paid for (GitHub no longer exposes the stargazer list, so there's no way to verify each one). Buying stars isn't about fame here; it's for "credibility" with SEO and AI crawlers. Search "卡券货源 API github" or "卡券 开放平台 API 签名下单 自动发货" and it's the first result on Brave. What's actually worth studying is the playbook: a GitHub repo used as a B2B acquisition page, with the README written for AI agents to read. The project itself hasn't taken off.

1. What it is

  • Repo: ksjhk/api, created 2026-10-26, MIT. Contents are only the README, docs/api.md, and example clients in four languages, size 26KB (GitHub API). There are only 13 commits: 10-26 pushed 10 commits in one go, and 11-08/09 added three commits for "public query + quick-start."
  • The business behind it: kashijie.com ("Kashijie Card World") recruits "voucher promotion agents" as a side hustle. The homepage reads "first-hand voucher supply, high commissions with instant settlement, easily expand your downline, earn while you recruit agents," plus "10000+ agent students community" and "live mentors guide you to your first sale" (official site text, read 2026-11-13). The open platform settles from a balance: agents prepay, then call the API to place orders, and the platform deducts at cost price. Each key can set a "markup multiplier" of 1.00~99.99, which agents use to add a uniform markup for their own downstream buyers.
  • Supply scale: the public endpoint /open-api/v1/goods returns total=123 products (QQ membership, Jimeng, iQIYI, Fengchao, Ele.me red packets, JD Plus, etc.), which does not match the README's "200+ brands" (2026-11-13 in our test). Product images come from cdn.ksjhaoka.com, the CDN of Kashijie phone-number cards (a traffic-card distribution platform under Fujian Rongyifei Technology), so it belongs to the same system. The GitHub account bio reads "Kashijie SIM official," but beyond the official SIM-card site there is a large number of agent-built "Kashijie" sites (kashijie.org, kashijie.online, ka-shijie.cn, ksjhk.online, etc.). ksjhk.online shares a name with this GitHub account, but whether the account is the company's official one or a large agent is unverified.

2. Growth data

DatestarDaily gain
11-0723—
11-0827+4
11-0935+8
11-1053+18
11-1177+24
11-12124+47
11-13148+24
11-14183+35

(From our GitHub repo_day; live GitHub API 11-13 reads 193.) 0 forks, 0 watchers (subscribers), 0 issues. The author has 1 followers; the account registered 2026-05-11.

3. Where the stars came from: the evidence chain

GitHub now returns 404 for every repo's stargazers endpoint (REST, GraphQL and the web alike, even for normal repos like glisse), so the accounts that starred cannot be inspected. The indirect evidence points to "most likely not organic," but this is unconfirmed:

  1. The same author has 11 repos, and only this one has stars. The more substantial, working ksjhk/kaquan (Laravel backend + static frontend + DSH plugin, 1.8MB) has only 1 stars. hotel-reseller and movie-ticket-reseller, created 11-09, are the same doc template with different keywords, and each has 0 stars.
  2. The author's own promo post pushed kaquan, not api. On Juejin 11-09 posted "Can't get membership card supply? I open-sourced a distribution platform: 200+ brand API, auto-delivery," with the link github.com/ksjhk/kaquan at the end and a "if you find it useful, give it a Star" call. Kaquan still has 1 stars, while the unpromoted api gained +160 over the same period. If the stars came from readers, they would not be in this direction.
  3. 0 fork 加 0 watcher。 只看 0 fork 不够说明问题:我们库里建库 ≤30 天、150–250 star 的 968 个库,有 358 个(37%)也是 0 fork。但连作者自己都没 watch(subscribers=0),也没有一个 issue,这种组合在真开发者项目里少见。
  4. Around the same time, a whole batch of similar Chinese commercial repos look exactly alike. In our database, repos created ≤45 days ago with 100–600 stars, 0 forks and Chinese descriptions, the ones gaining the most stars in the last ~7 days include the CoreGold virtual credit card matrix (a dozen different accounts, one brand, each around 100–120 stars, +7~38 every 45 days), laoma528/GPTplusSub (ChatGPT top-up tutorial, 182 stars / 0 forks), sunreview/anti-claude (157 / 0), and others. ksjhk/api ranks 3th in this batch. What these repos share: they run a business (top-ups, virtual cards, vouchers) and use GitHub as a landing page, with steady star curves and zero forks.
  5. Unrelated to the DSH plugin ecosystem. The recall rationale claimed it rode on the DeepSeek Harness plugin directory. In fact, the api repo has no topics, and its README never mentions DSH. The DSH plugin appears only in kaquan's README. Searching awesome-dsh-plugin.com's plugins.json and index.json (the index read by DSH's in-app Discover) turns up nothing for ksjhk, kashijie or Kashijie 2026-11-13 (our test), 0 entries in total. So this channel brought no stars.
  6. No spread found on social media or in the press. x-radar searches for "kashijie" and "Kashijie" returned 60 posts in the last ~0 days; no Hacker News appearance, no media coverage. The repo is less than 60 days old and would not have made trending anyway.

4. Growth levers (how users arrive)

ChannelEvidenceConfirmed
GitHub stars as "credibility"See section 3: 0 forks / 0 watchers, the author's other repos all at 0, the promoted kaquan did not grow, and a batch of similar commercial repos rose at the same timeNo (stargazer lists cannot be accessed)
Search engine SEO (borrowing GitHub's domain authority)Brave searches "card goods supplier API github" and "voucher open platform API signed orders auto-delivery" both return github.com/ksjhk/api as result 1, with kaquan at 2 or 3; "member card voucher API call example PHP Python" puts it at 2, and result 1 is its own Juejin article (2026-11-13)Yes
Tech-community sponsored contentJuejin posts 11-09, wrapping supplier integration in an "I open-sourced it" toneYes
Q&A seedingThe search summary of the Zhihu question "Is there a reliable virtual card and voucher supply platform?" shows www.kashijie.comYes (only the summary was seen)
Integration docs written for AI agents (GEO)The README's quick-start says "have the user provide an unregistered phone number; the program calls /platform/quick-start to register automatically and generate the first key pair." Submitting a phone number opens an account and issues keys immediately. The same author also built KSJAPI, ksjskill ("packaged as natural-language callable AI capabilities"), copyksjvip (a one-click voucher-site clone skill for Kimi), and the DSH plugin in kaquanYes (built); effect unknown
Agent recruitment and community mentoringkashijie.com: "easily expand your downline, earn while you recruit agents" and "10000+ agent students"; the recall card says the FAQ describes unlimited-level downline commissionsYes (self-reported on the official site; figures not verified)
Keyword-matrix of repos11-09 built hotel-reseller and movie-ticket-reseller at the same time, with the same doc template aimed at the search terms "hotel booking supply" and "movie ticket supply"Yes

Where the money comes back: agents first top up the open-platform balance (WeChat / Alipay QR codes), the platform deducts at cost, and the gap between cost and the brand's official price is the margin. For example, QQ membership monthly card cost 9.93 yuan, face value 14 yuan; Jimeng premium annual card cost 4962.75 yuan, face value 10398 yuan (public endpoint). Agents then resell at their markup multiplier to downstream buyers, and the platform can take a cut of downline commissions. Stars only make potential agents feel, in search results and on GitHub, that "this is a legit open-source project."

5. Competitors

NameWhereLaunchedScaleHow it differs
Fulu Open Platform / Fulu Juicai (fulu.com)WebEarly (Fulu Holdings, HK-listed company)Claims to connect 1,500+ brands and 2,300+ scenarios, with about 22,000 productsRequires company documents, signed agreements and review to join; Kashijie lets you check goods without registering, and submitting a phone number opens an account
Lekayi (lekayi.com)WebNot verifiedClaims 3,000+ SKUs, B2B, OpenAPITargets enterprise procurement; Kashijie targets individual agents and side-hustle people
Lechen Shuka, XC Rights Shuka, Su Rights, Laohuguа, Rights Shuka (mcsk / vxes), Orange CouponsWebNot verifiedSmall and mid-size supply terminals, all advertising "free API integration," relying on sponsored-article sites for SEO (e.g. tolc.cn)Same SEO sponsored-article playbook, but none has a real repo on GitHub; Kashijie got there first with the "GitHub + developer docs" entry point
assimon/dujiaoka (Dujiaoka card-issuing)Open sourceAbout 201812,125 stars (our database)A card-issuing storefront system that brings no supply; kaquan is a bundle of "storefront + in-house supply"
lizhipay/acg-faka (Acg-faka card-issuing)Open sourceAbout 2021-105,911 starSame as above: system only, no product sales
dimthink/PriceAIOpen source2026-05-085,457 stars, 248 forks, 6 watchersAn AI-subscription card-marketplace price comparison in an adjacent niche; a fork rate of 4.5% is normal, so it serves as a control for organic growth
cd-zwj/kamihub, jkankan51-byte/kaquan-nav, xwdeshenghuo/rights-price-hubOpen source2026-09~112 / 0 / 0 starGitHub landing pages in the same niche, but without bought stars, so they did not rank
Kashijie SIM cards (ksjhaoka)Web / app (Tencent App Treasure)About 2022Phone-card distribution platform with many agent sitesSame brand system, phone-card niche; vouchers are a newly added category

Conclusion: the card-voucher supply API market is crowded, with dozens of supply terminals using nearly identical wording. On GitHub, this is the only card-voucher supply repo with ≥50 stars (our database searched by homepage domain and description keywords; no comparable repo found).

6. Differentiation

  • Product layer: read-only endpoints are publicly accessible without signing up (you can compare goods without registering); submitting a phone number opens an account and issues keys automatically; each key carries its own markup multiplier; vouchers, hotels and movie tickets are covered together. For individual and side-hustle agents, the entry bar is much lower than for enterprise-onboarding players like Fulu. The Juejin article also features these points.
  • Channel layer: other supply terminals write sponsored articles on Baidu and Zhihu, while this one added a "GitHub repo + bought stars" route, borrowing github.com's domain authority to rank for developer search terms like "voucher API." The README is also written as a flow an AI agent can follow directly (get phone number → call API → get key), betting that people will one day ask Cursor / Claude / DSH-type tools to "connect a voucher supplier" for them.
  • Is that why it took off?: the stars were produced by a channel move, not by product differentiation. Whether the product differentiation keeps agents around cannot be seen in the data: kaquan has 1 stars and 0 forks, and there is no third-party discussion of any kind.

7. Visibility and copycats

  • Visibility: almost nobody knows about it yet. No X, HN, or press coverage. The only mentions I could find are its own Juejin posts and Zhihu sponsored content. Revenue isn't public (and that's not really how open-source projects work).
  • Copycats: nobody is copying it. Similar projects in the same niche like kamihub, kaquan-nav, and rights-price-hub popped up in parallel, and none of them gamed their stars. Zooming out, "Chinese business repo with inflated stars plus landing page" is already a pattern (CoreGold's dozen-plus repo network, GPT top-up tutorial repos, etc.). This one is part of that wave, not the starting point.

8. What others can learn

  1. For B2B supply and API-style businesses, a GitHub repo can serve as a second homepage. github.com has very high domain authority, and a README with multilingual examples can quickly rank for long-tail queries like "X supply API" or "X API call example". One repo per keyword (hotel-reseller, movie-ticket-reseller) costs almost nothing.
  2. Write the README for AI agents. Compress "sign up → get key → call" into steps a machine can follow, with no manual review and no slider captchas. Then add a skill or plugin package (DSH, Kimi skill) and wait for coding agents to integrate it on people's behalf.
  3. Costs and risks: buying stars violates GitHub's terms and risks having stars purged or the account banned. Experienced developers get suspicious when they see 0 forks / 0 watchers. And the card-resale "top-up" side carries its own platform risk-control and compliance issues.

9. Uncertain

  • Whether the stars were actually bought: GitHub doesn't publish stargazer lists, so the evidence is indirect.
  • Whether this GitHub account is the official account of Kashijie Company or a large agency (same name as ksjhk.online): unverified.
  • How many sign-ups and top-ups this playbook actually brings in: not visible. kashijie.com's traffic isn't in the data (it's not in Tranco's top 1M), and neither are registration numbers.
  • "Unlimited-level commissions" and "10000+ agent students" come from the official site and recruitment cards. Not verified.

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