Stringers
ZiChuanLan/freebie-radar · ZiChuanLan

A "free-stuff radar" got 860 stars in 4 days: the author's reputation built up on L 站, plus one repost with 150k views.

What it is A self-hosted aggregator for free-stuff alerts. It pulls "free claim" posts from 60+ forums, game platforms and Telegram channels, uses AI to separate truly free items from lotteries, trials and promos, and pushes them to Telegram.

Why it blew up The author already had a name on LINUX DO. The launch post earned the first batch of stars; the next day a Chinese resource account with 144k followers posted it, pulling in 151k views and 2,500 bookmarks and sending it viral. It also landed as L 站 was flooded with AI-quota freebies that were hard to tell apart from scams.

CHow you could follow it Don't build a third free-stuff radar. Borrow the playbook: multi-source aggregation + AI filtering + push notifications + a public demo site, applied to another scattered, hard-to-verify niche. Launch in a vertical community that plays by its rules first, then get a Chinese resource account to repost it.

Grade C: within a week there were already two projects on the same topic. The winner rode the author's community reputation and one hard-to-copy big-account repost; followers will most likely get a few hundred stars at best. The playbook is worth learning, the topic isn't worth building again. Circulating in small circles
Data (updated daily) · repo created 2026-11-12 · 1,785 stars
GitHub · stars
01k2k11-1511-1611-1711-1811-1911-201.8k
Data 2026-11-15 → 2026-11-20 (6 days, 6 observations). Points before 10-28 come from deeply / archive backfill, not every day has one

A Chinese developer spent 5 days writing an open-source tool that automatically gathers "free claim" posts from 60+ forums, game platforms and Telegram channels, then uses AI to filter out lotteries, trials and promos. It hit 860 stars in 4 days. In the same week, a near-identical project stalled at 500. The gap between them was mainly distribution, not code. Below: how this playbook ran, which parts you can borrow, and why I wouldn't recommend building another one on the same topic.

A forum post at 2 a.m.

At 17:2 AM Beijing time on Nov 14, 2026, a user called 岚 (ZiChuanLan) posted on LINUX DO (what devs in the community call the "L Site") with a very long title: "[Resource Radar] A self-hosted aggregator that filters free-for-all deals, AI resources, redemption codes, games, freebies and news from across the web. Includes 60+ sources. Mom, you never have to worry about missing free stuff again!"

The post opened with the open-source promo disclaimer L 站 requires: fully open source, with an AI-written intro shown in screenshots. The body was just a few UI screenshots, a demo site link, and one line: "Don't want to deploy it or spend tokens? Just check out my site." 14 minutes after posting, the repo got its first fork.

By 11-16 the post had 1,900 views and 170 likes and made L 站's weekly top list. The most-liked reply was "Can you make a Cloudflare version? If we're freeloading, let's go all the way" (13 likes), followed by someone asking whether there was a "generous version," meaning a public instance anyone could use. The author replied the same day. Later someone opened a PR that produced a Cloudflare Workers version, and the README added a public multi-user mode that supports L 站 account login.

The author isn't new. Their account carries L 站's "mentor" title, and in October they had posted two projects, agent-purge and pi-purge, which got 1,835 and 794 stars. They have only 24 followers on GitHub; their reputation is almost entirely on L 站.

What it is

A self-hosted "freebies" aggregator. It pulls from 63 sources on a schedule, 33 of them enabled by default: Chinese tech forums like L站, NodeSeek, and V2EX, free game giveaways from Epic, Steam, and GamerPower, free AI models on OpenRouter, plus Reddit, Telegram channels, and overseas deal sites. Each item first goes through keyword rules, then AI tags it: is it actually free, what category it falls under, and how to get it (claim directly, conditional, sweepstakes, trial, promotion, or paid). Finally it pushes to Telegram according to the rules you set.

It runs with a single docker compose up, no separate database to install. MIT licensed, free. The author's demo site radar.zichuanlan.top has been running nonstop: 11-16 the database holds about 10k entries, roughly 5,900 new in the last 24 hours, with the AI keeping 6,222 and rejecting 2,569 (public stats from the demo site's API).

The code was written over 5 days in 106 commits, and the commit log includes messages like "third-round audit fixes," suggesting it was rushed out with heavy AI assistance (inferred).

Where the stars came from: a two-step relay

DateDaily new starsWhat happened
11-13About 0L 站 post, evening of 18:17 (UTC)
11-14+184"Weekly resources" recommendation post, 11:53 (UTC)
11-15+472The repost kept spreading, and reworded copies followed
11-16+51Decline

(Trendshift daily numbers; GitHub live count as of 11-16: 860 stars, 145 forks)

Step one was the L 站 post, which brought the first batch of stars and real demand. Step two did the heavy lifting: on the evening of 11-14, "Weekly Resources" (144k followers on X, bio: "sharing AI materials, books, useful tools, side hustles and more") posted: "Holy cow! Someone turned every free-stuff source into an automatic monitoring radar!" with the line "Best of all, it plugs in AI to filter automatically and tell real deals from fake," and the demo site and GitHub links at the end. It reached 11-16 with 151k views, 1,800 likes, 2,500 bookmarks and 230 reposts.

Bookmarks outnumber likes, so people are saving it as a tool. Forks peaked at 11-14 around 15–17 PM (UTC), 5–8 an hour, just hours after the post went up.

Then came a chain of reposts: another X account 11-15 reposted it word for word, forums ran the full text, Threads has auto-repost accounts, and a Russian YouTube Shorts video titled "Top 10 GitHub projects of the past three days" included it. HN, Reddit and media didn't cover it, and I couldn't find the author's own post on V2EX or NodeSeek. Under the 周览资源 post someone asked "You made this yourself?" There's no evidence the recommendation was organic versus paid (unverified).

Another freebie radar from the same week, stalled at 500

luobinger/PerkRadar launched 4 days before it, with nearly the same topic and the same stack. GitHubDaily posted it on X once, that post got about 3,500 views, and the project peaked at 183 stars in a day before flattening out. It now has 510 stars.

Put the two side by side and the differences are clear:

PerkRadarfreebie-radar
Sources763
CategoryClassified by title rulesAI checks for fakes, sorts by "how to get it"
DeliveryNoneMulti-channel Telegram push
Public demo siteNoneYes, usable without deploying
LaunchedPosted on NodeSeekAuthor has a reputation on L站
Biggest exposureAbout 3,500 views151k views

The product differences are real, and they show up in the copy people repost ("63 data sources", "tells real perks from fake ones"). But exposure differs by 40 times, while stars differ by only 1.7 times. Stars mostly come from distribution. The better product just converted the same exposure into more stars.

Why this week

This year L站's freebie section has been flooded with AI-quota posts: public-service API stations, model invite codes, free GPT access. In the L站 weekly chart the same week, a single invite-code roundup got 14k views. There are so many freebies nobody can keep up, and mixed in are plenty of giveaways, promos and fake deals. "AI filters out the fakes for you" speaks directly to that pain.

The people who care about freebies are the same people who reshare this kind of thing: L站 users and the "freebie hunter" crowd on Chinese X, and those repost accounts themselves post freebie resources. For them, a tool that helps people hunt freebies is ready-made content.

The topic itself isn't new. Chinese deal-tracking sites and e-commerce bargain sites have run for years, and L站 and NodeSeek already had Telegram keyword-monitoring bots; the L站 one claims to have pushed 1.84M messages. The format isn't new either: sansan0/TrendRadar uses "multi-source aggregation + AI filtering + push" for hot-search monitoring and has 63k stars. freebie-radar takes that proven format and applies it to a topic that's easier to bookmark and share.

The gap it left

  • The scope is widening. The README now calls it "资源雷达 · Resource Radar" and covers Black Friday VPS deals, group buys, secondhand goods and news. Of the demo site's three largest sources, a Chinese deals Telegram channel, Slickdeals and 什么值得买, only 1,645 of 10k entries the AI tagged as "freebie". It's drifting toward a general deals aggregator, and its original "true freebies only" pitch is getting diluted.
  • Dependence on L站 is a risk. One L站 thread asked "Is scraping L站 against the rules, is there a risk of getting banned?" Under the X post, someone also warned that scraping sites like L站, which run Cloudflare protection, needs a separate proxy or FlareSolverr.
  • Momentum is fading. 11-16 only gained 51. The author's previous two projects followed the same pattern: a spike for a day or two, then back to a few dozen stars a day.

If it were you

Verdict: C. You can learn from it, but building another one on the same topic isn't worth it. Within a week there are already two freebie radars. The winner relied on the author's reputation on L站 plus a repost that could have reached 150k views. If you build a third, the best realistic outcome is probably PerkRadar-level, around 500 stars (estimate).

Who it suits: developers who want stars for their open-source project and already have some standing in a niche community (L站, NodeSeek, V2EX, or an English subreddit or Discord), and who can use AI to ship a deployable tool with a live demo within a week.

Who it doesn't suit: people with no account history in any community who expect GitHub itself to bring in users. A new repo needs roughly 60 days before it can reach GitHub trending, so these projects' stars depend entirely on posts. And translating it into English and reposting won't work either, since it already has an English README.

How to do it. What's worth borrowing is the playbook, not the topic:

  1. Change the topic, keep the format: "multi-source aggregation + AI filtering + push + public demo site", applied to a subject where information is equally scattered, real and fake are equally hard to tell apart, and readers would want to bookmark it. For example: free credits and student discounts useful to developers (English-language community), remote work and freelance gigs, or limited-time free and discounted software (estimates; I haven't verified whether ready-made projects exist for each). Search GitHub first, and don't let someone else ship the same thing in the same week.
  2. Fill the gap it left: a narrower version that covers only "true freebies" without deals or news, or an English version focused on overseas free resources, with sources and judgment rules rebuilt for English communities.
  3. Build an audience before the project: the author shipped 3 repos in 4 weeks with the same playbook, totaling about 3,500 stars. Build a name first by shipping one or two small projects in a single community. The third project is the one you're aiming for.

How to get users: follow the two steps freebie-radar took. First, post an open-source promotion in your niche community following its rules, with screenshots and the demo link, and respond to feature requests the same day, shipping fixes the same day. Second, put one selling point that can be pasted straight into a tweet at the top of the README (theirs was "63 sources + AI identifies real freebies"), then reach out to large Chinese X accounts that share tools and resources, because they're short on exactly this kind of material. If people bookmark more than they like, you've hit the mark.

Time window: there's no clear deadline for the topic-switch route. But the wave of AI-quota freebie hype on L站 is still going. If your topic sits next to freebies or free AI credits, it's probably worth starting this year (estimate).

Signs I got this wrong: (1) if freebie-radar is still gaining a hundred or two hundred stars a day after a month or two, the ceiling for this topic is higher than I estimated, and a narrower or English version would be worth trying; (2) if a third or fourth freebie radar appears and also clears 500 stars, demand matters more than distribution, and people without a reputation still have a shot; (3) conversely, if same-format projects on new topics can't get off the ground, then this one relied mainly on the author personally and the repost, and others can't borrow that.

Data and sources

GitHub: ZiChuanLan/freebie-radar repo page, README, commit and fork history (11-16 checked live); luobinger/PerkRadar, sansan0/TrendRadar and the author's other repos' star counts. Trendshift: daily new stars. LINUX DO: the original topic/3173875 post and replies (posted 11-13). X: @grgerwcwetwet 11-14 recommendation post and comments, @GitHubDaily's post about PerkRadar. Demo site radar.zichuanlan.top public stats endpoint (11-16). Where "estimate" or "unverified" appears, that's an estimate or an unverified claim.

GitHub repo ↗

Research notes: competitor table, acquisition evidence, full research report
Competitors
NameWhereLaunchedScaleHow it differs from this one
luobinger/PerkRadarOpen source2026-11-08510★; GitHubDaily X post 3.5K viewsSame topic and stack, but only 7 sources, title-rule classification, no push; no big-account repost, stalled at 500★
Xianbao Cool (xianbao.fun)Web2018Long-running Chinese deal aggregatorCovers e-commerce and lifestyle deals; centralized and not self-hostable; doesn't cover L站/NS AI-quota or VPS-style geek freebies
@Linuxdo_keyword_bot / @ns_keyword_bot and other keyword-monitoring botsOtherFrom ¥2025The L站 bot claims 1.84M messages pushed in totalSingle-site keyword matching; no cross-source aggregation, no real/fake judgment
sansan0/TrendRadarOpen source202563.6K★Same "multi-source aggregation + AI filtering + push" template, applied to hot-search and public-opinion monitoring rather than freebies
GamerPower / FreeStuff Bot / Epic Freebie Games RadarOtherYearsLarge Discord bots and standalone appsOnly game giveaways; freebie-radar uses them as upstream sources
什么值得买 / 线报屋WebYearsBigE-commerce deals + cashback, not aimed at developer-community freebies
abinnz/xiaofendui, 6dylan6/jdmOpen source2018 / 2022466★ / 338★Single-channel WeChat / Qinglong deal scripts, no longer maintained
panxunying/ai-coding-welfare, free-model-radarOpen sourceFrom ¥2026-081,334★ / unverifiedOnly covers AI quotas / public-service API stations and similar
RSSHub + RSS-to-Telegram / FoloOpen sourceYearsBigYou have to piece together subscriptions and filtering yourself, with no real/fake freebie judgment
Where users come from
  • Verified LINUX DO open-source promotion post: linux.do/t/topic/3173875,11-13 posted 18:17 UTC, 1,924 views, 170 likes, 43 replies, entered the L站 weekly chart; the first fork appeared 14 minutes later
  • Verified Chinese X big-account repost: @grgerwcwetwet (周览资源, 144k followers) 11-14 11:53 UTC recommendation post with 151k views, 1.8K likes, 2.5K bookmarks; afterward Trendshift single-day star gains rose from +184 to +472, with forks peaking at 11-14 15–17 UTC
  • Verified Secondary reposts / trend sites: @0x8888 reworded repost (11-15), bbs1.org reprint, Threads auto-repost, Russian YouTube Shorts "Топ-10 GitHub за 3 дня", listed on Trendshift / RisingTech / GitHunt
  • Verified Public demo site: radar.zichuanlan.top is live, with about 10k entries and 24 new in the last 5,924 hours; the post says "if you don't want to deploy, check out my site"
  • Verified Author's existing reputation and repeated playbook: October's pi-purge 794★, agent-purge 1,835★, both launched on L站 with reposts from Chinese X accounts (e.g., @LiuweijiaVip)
  • Inferred Paid promotion / mutual shoutouts: someone asked "You made this yourself?" under the 周览资源 post; no evidence
  • Inferred V2EX / NodeSeek / HN / Reddit / media: searches turned up no posts or coverage by the author

Why ZiChuanLan/freebie-radar (资源雷达 / 白嫖雷达) took off

Bottom line first

freebie-radar is a self-hosted "free stuff / deals" aggregator: it crawls 63 sources on a schedule (LINUX DO, NodeSeek, V2EX, Epic, Steam, GamerPower, Reddit, Telegram channels, and more), uses rules plus AI to tell real freebies from fake ones, keeps giveaways, trials and promotions in separate buckets, and pushes matches to Telegram by rule. The author got 4 stars in 858 days. The source is clear: two steps in relay.

  1. 11-13 18:17 UTC, the author posted an open-source promo thread on LINUX DO. The author already had a name on L站: earlier projects pi-purge and agent-purge both blew up there, and the account carries a "mentor" title. The thread got 1.9K views and 170 likes and made it onto L站's weekly chart. 14 minutes later came the first fork.
  2. 11-14 11:53 UTC, the Chinese resource-repost account "周览资源" (@grgerwcwetwet, 144k followers) recommended it. The post got 151k views, 1.8K likes and 2.5K bookmarks. Trendshift's daily star gain jumped from +184 at 11-14 to +472 at 11-15; in our database, stars went 11-15→11-16, a +508 single-day gain.

It hit a real trend: over the past year, "AI credits / public API relays / invite codes" freebies on LINUX DO exploded, the info was scattered across a dozen forums and channels, and telling real from fake was hard. It is the first open-source tool to put "Chinese community freebies + free game giveaways + free AI credits" in one site and use AI to filter out fake offers, and it shipped with a working public demo site. The topic itself is old: tip sites and keyword-monitor bots have existed for years. So the main driver was the author's community credibility plus reposting by big accounts; the product's differentiation was secondary.

Key facts

itemsNumbersSource
Repo created2026-11-12 09:08 UTC (first commit 11-11 01:11 UTC)GitHub API
star / fork858 / 145 (checked 11-16); card 773 / 126GitHub API; our github database
Daily11-15 snapshot 265 → 11-16 773 (+508)our GitHub repo_day
Trendshift daily11-13 +0, 11-14 +184, 11-15 +472, 11-16 +51trendshift.io/repositories/356297
Fork timing distributionFirst fork 11-13 18:31 UTC; 11-14 15–17 UTC, 5–8 per hour, the densest stretchGitHub forks API
commit106, finished within 5 days (commits like "third round of audit fixes" and "impeccable design spec" point to heavy AI-assisted development)GitHub commits
External PRs / issuesSomeone built a Cloudflare Workers port (answering the L站 request "can you make a CF version?"), a docker fix, and a Telegram push fixGitHub issues/PR
AuthorThe GitHub account has only 24 followers, but agent-purge has 1,835★, pi-purge 794★, some-magic 208★GitHub API
Demo siteradar.zichuanlan.top is live; /api/stats shows about 10k items, 24 in the last 5,924 hours, with AI keeping 6,173 and rejecting 2,554Live query

Note: the stargazer detail endpoint returns empty for this repo (REST 404, GraphQL an empty list), so we can't get per-star timestamps or accounts and can only infer indirectly whether stars were bought. Fork timing lines up with the two external exposures, and the external PRs come from real users, so it doesn't look botted.

Competitors: who else does the same thing

NameWhereLaunchedScaleHow it differs
Xianbao Cool (xianbao.fun)Web2018Long-running domestic tip aggregator with push notifications and filters (someone has asked about plugging in RSSHub)Aggregates e-commerce / lifestyle deals (Zhuanke Ba, Weibo, Douban). Centralized website; doesn't cover LINUX DO / NodeSeek's AI credits, VPS, or open-source geek freebies, and can't be self-hosted
Smzdm / Xianbao Wu and similarWeb / AppYearsBigMainly e-commerce deals with affiliate revenue; not developer-community freebies
@Linuxdo_keyword_bot, @ns_keyword_bot and other keyword-monitor botsTelegramFrom ¥2025The L站 bot claims 1.84M+ pushes; the NS bot 280k+Keyword matching on a single site only; no cross-source aggregation, no real-vs-fake judgment
luobinger/PerkRadarOpen source2026-11-08 (4 days early)510★; GitHubDaily X thread 3.5K views; NodeSeek postsNearly the same topic and stack (node:sqlite + dashboard), but only 7 sources, categorization by title rules only, no push. Peaked at +183 in a single day, then flatlined
sansan0/TrendRadarOpen source202563.6K★Same template (multi-platform aggregation + AI filtering + push), but built for trending topics / public sentiment, not freebies
GamerPower / FreeStuff Bot / Epic Freebie Games RadarWeb, Discord, iOS / AndroidYearsFreeStuff is a large Discord bot; Epic Freebie Games Radar has a standalone appOnly covers free game giveaways; freebie-radar uses GamerPower, Epic, Steam and CheapShark directly as sources
abinnz/xiaofendui, 6dylan6/jdm and other deal-tip scriptsOpen source2018 / 2022466★ / 338★WeChat bots or Qinglong scripts, single channel, long unmaintained
panxunying/ai-coding-welfare, a1667834841/free-model-radarOpen sourceFrom ¥2026-081,334★ / unverifiedOnly covers the AI-credits / public-relay navigation niche
RSSHub + RSS-to-Telegram, FoloOpen source / AppYearsBigYou have to build your own subscriptions and filters; no real-vs-fake judgment

Verdict: there's no shortage of competitors. This is the first open-source project to put Chinese geek-community freebies + games + AI credits + overseas freebies into one self-hostable workspace, and to use AI to grade how each item is obtained. The closest rival, PerkRadar, came 4 days earlier with nearly the same topic, but stalled at 500★; that comparison says the most (see below).

Growth: where users come from

  1. LINUX DO open-source promo thread (confirmed). linux.do/t/topic/3173875,11-13 was posted 18:17 UTC in the "Dev Tuning" category, tagged "open-source promo / freebies," and disclosed the AI-generated parts per community rules. 1,924 views, 170 likes, 43 replies, and it made L站's weekly chart. Requests in the thread (a CF version, a "public good-samaritan" public site, LINUX DO login) got a reply from the author the same day; the README later added SITE_MODE=public and a LINUX DO OAuth multi-user mode. The first fork came 14 minutes after the post.
  2. Big-account repost (confirmed, the main driver). 周览资源 @grgerwcwetwet (144k followers, bio: "sharing AI resources / tools / side-hustle info," also runs a TG channel) posted 11-14 11:53 UTC, opening with "Holy cow! Someone turned the whole web's freebie sources into an automatic monitoring radar!" It got 151k views, 1.8K likes, 2.5K bookmarks and 230 reposts. Bookmarks outnumber likes, meaning people saved it as a tool. A commenter asked "did you build it yourself?"; any partnership is unverified.
  3. Secondary repost chain (confirmed). @0x8888 reposted the same copy with a rewrite on 11-15 23:17 UTC; bbs1.org/topic/3348 carried the full text; the Threads account @dooo.sth.design auto-reposted it (at 698★); a Russian YouTube Short "Top-10 GitHub for 3 days" included it; Trendshift, RisingTech, GitHunt and other trend sites picked it up.
  4. The demo site is the product (confirmed). The post said "if you don't want to deploy or don't want to burn tokens, just check out my site." A public demo lets people who can't deploy use it right away, lowering the bar for starring.
  5. The author has reputation (confirmed). The same author, October, posted on L站 about pi-purge (a "armor-breaking" persona plugin ported from dsh-purge) and agent-purge (an update post, "1K stars!"), both using the "first launch on L站 → Chinese X account repost" pattern (e.g. @LiuweijiaVip, "GitHub viral open-source tool agent-purge"), reaching 794★ / 1,835★. This is the third reuse of the pattern.
  6. Not found: no posts by the author on V2EX or NodeSeek (Jina search); no HN, Reddit or media coverage; no ads. GitHub trending only admits repos about 60 days old, which doesn't apply here.

Why this one, and why now

  • Timing: L站's freebies section was in a boom. That same week, L站's weekly chart was filled with AI-credit freebies: the "Muse invite code" thread (14.5K views), "public relay gives $1000", "free GPT binge." There were so many freebies nobody could keep up, mixed with lots of promos, giveaways and fakes. "AI second-pass filtering of fakes" and "sorting giveaways / trials / promos into separate buckets" hit exactly that pain point, which is why the first line of the README says it is "never contaminated by paid deals, group buys, secondhand items or news."
  • The audience and the spreading channel are the same people. The people who care most about freebies are the "freebie hunters" on L站 and Chinese X, and the repost accounts' content is itself "free resources." A tool that "helps you get freebies for free" is a natural post for them; 2.5K bookmarks is the evidence.
  • High polish. One docker compose command gets it running, ships with 63 built-in sources, 10-route cover images, one-click copy for redemption codes, multi-route Telegram push, zero-token re-categorization, plus a live demo site. That matches what people expect from something like TrendRadar: "deploy it and it works."
  • Comparison with PerkRadar. Topic, timing and stack are nearly identical. PerkRadar was also featured on GitHubDaily, but that post got only 3.5K views, and it stopped at 500★. freebie-radar added three things: the author's L站 credibility, a viral repost with 150k views, and sharper selling points (AI fact-checking, 60+ sources, Telegram push, a public site). It suggests the gap was mostly distribution scale; the product differences amplified it.

Differentiation

  • It differs from others in three ways: ① source mix: Chinese geek communities (L站, NS, V2EX), free game giveaways, free AI credits (OpenRouter :free) and overseas freebies (Hip2Save, Reddit) in one place, a combination neither tip sites nor single-site bots have offered; ② a three-axis classification of format × domain × how to obtain it, where AI separates giveaways, trials, promos and paid items from truly free ones; ③ self-hostable, with a public site you can browse directly, and multiple users who can each bind their own Telegram.
  • These differences are the hooks in the promo copy ("the best part is it has built-in AI filtering that can tell real freebies from fake ones"), so they did contribute to the breakout. But the decisive factor was distribution: PerkRadar, a close peer, had no big-account reposts and stalled at 500★.

Is anyone copying it?

  • Someone has already deployed it as a public site: 2bll.com, titled "白嫖雷达 · Freebie Radar," running an older pre-rename version.
  • Community port: someone submitted a PR for a Cloudflare Workers + Durable Object version.
  • Rewritten reposts: @0x8888, bbs1.org.
  • No independent knockoff spotted yet; PerkRadar came earlier and is a parallel competitor, not a copy.

Visibility

It's circulating in small circles: one L站 thread (1.9K views) plus a Chinese X repost (150k views) and its rewrite chain. No media coverage, no HN or Reddit, and nothing in the English-speaking world. MIT licensed, no pricing, no revenue.

What others can borrow

  • Two-step community launch plus big-account repost: first post an open-source promo in a vertical community (L站) following its rules to get the first 100–200★ and real feedback, then get carried further by Chinese X resource-repost accounts. Within 4 weeks this author used the same playbook to ship 3 repos, about 3,500★ combined.
  • Swapping the vertical onto a proven template: TrendRadar showed the "multi-source aggregation + AI filtering + push" shape works, and this project moved it to freebies, a topic that drives bookmarks and reposts.
  • The public demo site is itself the product: people can use it without deploying, which converts more stars.
  • Monetization room: the demo site already has multi-user accounts, personal RSS and Telegram binding. The next step would be a paid push channel or affiliate / promo slots (the classic tip-site business model). The author hasn't done this yet.

Still unclear

  • Whether 周览资源's recommendation was organic or paid / a mutual promotion: a commenter raised the question; there's no evidence either way.
  • No stargazer breakdown available, so account quality can't be verified.
  • The demo site's real traffic (new domain, not on Tranco).
  • Will the heat last? 11-16 Trendshift shows only +51. The first two repos, pi-purge and agent-purge, spiked for a day or two and then fell back to a few dozen stars a day.

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