Stringers
yt1988/who-are-you · yt1988

"Running a Japanese node, yet the site tells at a glance that I'm Chinese": a fingerprint self-check page, 9 days, 2M visits, earning only 400 stars

What it is An open-source web page that, as soon as you open it, reads out the hundreds of items a website can see without any permission, and assembles them into a portrait like "you're probably Chinese, and you're in Tokyo right now," then points out which spots give you away when you're behind a proxy and how to fix each one.

Why it spread During the Claude account-ban wave, Chinese users were worried about whether their setup was clean. The page checks timezone, DNS, WebRTC, and Claude's exit node all in one go. The result is a funny identity mismatch that's easy to screenshot and share. The author launched on V2EX and NodeSeek the same day under a hook headline, and a Chinese X post with 109K views picked it up.

CHow you could follow it Stop building yet another Chinese fingerprint-check site. Borrow the format instead: turn a detection tool's readings into identity-flavored plain language, add a grade, a fix, and a privacy-free share card. To win stars in this space for real, build what a web page can't reach, such as desktop or command-line environment self-checks, and one-click fix rules and plugins.

Grade C: its stars are a side effect of 2M web visits, driven by a one-time account-ban wave and the author's marketing. The Chinese proxy-tool niche is already crowded, and the two repos that followed 11-08 have only 0 to 1 stars, so copying it won't get you stars. It's been circulating in small circles.
Data (updated daily) · Repo created 2026-11-01 · 773 stars · 7 days +138
GitHub · stars
025050075011-0311-0711-1111-1611-20773
Data 2026-11-03 → 2026-11-20 (18 days, 18 observations). Points before 10-28 come from deeply / archive backfill, not every day has one

On Nov 2, 2026 at 10:53 a.m., a GitHub user named yt1988 posted on V2EX under the title "Running a Japan node, the site spots at a glance that I'm Chinese. So I built a tool to see who sold me out." The same post went up on NodeSeek that day. The page it links, who.1ms.dog, opens with a line: "You might be Chinese, and right now you're in Tokyo." Nine days later, the author said the page had about 2M visits, yet its open-source repo had only 425 stars. Browser fingerprint detection is a decade-old category. How did this one break through? Why is there such a gap between traffic and stars? If you want stars on a similar project, which half should you copy? This piece lays it out.

What the post said

The opening is a self-deprecating joke:

A Japan node, the IP checks out as Tokyo, an Amazon data center. I thought the site would never recognize me now. ...The IP changed, but the timezone didn't, the browser language didn't, and DNS still pointed to China. The worst was WebRTC: it reported my home Unicom IPv6 outright. The proxy was running for nothing.

Here's what it actually is: a page that asks for no permissions. It reads everything a website can already read and stitches it into a picture of "you, as the site sees you." On the right it lists the give-aways: timezone and IP don't match, DNS is in China, WebRTC leaks, data-center IP, UA claims a Mac but the GPU is NVIDIA. Next to each one is a "how to fix it." The author says he fixed his own setup and dropped from 7 to 2 exposures. The post ends with a hook: "Reply and tell me how many ways your proxy leaks."

The V2EX thread drew about 6,600 views and over 70 replies in 23 hours. Most people were counting their own leaks: "True, 6 exposures" "Two: timezone and fingerprint" "Tested as pure Aryan, shows nothing." Someone suspected "these sites may be phishing for enforcement, then selling data to security firms." The author replied that the code is open source and audits are welcome. One comment captured the context: "The Claude bans made everyone take browser fingerprinting seriously."

What it is

An MIT-licensed web page that runs on Cloudflare Workers. npx wrangler deploy gives you your own copy, and it supports 10 languages. It now runs 124 checks on load (the launch post said 71). On the server side it looks at IP, ASN, TLS/JA4, HTTP/2, QUIC and TCP fingerprints. On the browser side it looks at timezone, language, GPU, fonts, Canvas and audio fingerprints, and installed plugins. It also checks whether the exit looks the same to domestic sites, foreign sites, chatgpt.com and claude.ai.

Results come in three parts: a personified portrait; 14 "disguise checks," each graded A to F with a fix (how to set Clash's fake-ip / TUN, how to turn off WebRTC, where to change timezone); and a shareable card that leaves out IP and fingerprint by default. The city is included only if you tick it. Friends who open it see your portrait and a button reading "Test me." The page also saves your last result locally and tells you "Last time 8 exposures, this time 7."

In 9 days the repo got 154 commits. After the launch post, checks grew from 71 to 124, and disguise checks grew from 9 to 14.

The author describes himself as a former marketer and former media worker with "0 coding background," who wrote it line by line with AI on a Claude subscription. This page is one of the tools on his 1ms.dog site. The same site has more than a dozen sub-sites for speed tests, IP quality, DNS, streaming and AI unlock checks, each with its own repo.

Why now

The Claude ban wave. From late October to early November 2026, bans and KYC checks on Chinese users were at their peak. Huxiu ran "A Brief History of Claude Bans" on 10-28, 36Kr's 36 11-03 reposted "Buying Bans," and NodeSeek was full of "KYC twice in one week" posts. The anti-ban checklists circulating in the community treat clean DNS, matching timezone and language to the node, and a clean browser fingerprint as standard. This page runs all of them at once and even tests what claude.ai sees. In the same period, ping0.cc, which only checks IP cleanliness, climbed from 67k to 41k in global site rankings, showing demand for "environment checks" rising overall. This page caught that wave.

A paragraph travels better than a table. Older tools like BrowserLeaks and CreepJS list hundreds of data points in tables that only engineers can read. "You might be Chinese, and right now you're in Tokyo" is a mismatch of identity that's funny in itself. On linux.do someone posted "Found a site that profiles you. I came out as a super hybrid," and at least 6 NodeSeek users started threads to show off their results. One was titled "My Claude should be invincible now," and it ran 4 pages.

Test, fix, test again. Saying "you're exposed" gets a glance and a click-away. This tool gives a fix for each system, a grade, and remembers your last result, so users change their config and come back to test again. The first feature request in the V2EX comments was "add a one-click fix button next to each exposure."

The launch was marketer playbook. A hook title, an opening with personal history, and a question at the end to drive replies. Posted the same day to NodeSeek and V2EX, plus the author's own Telegram channel. On day four came a second post: "How I built a tool site with 2M visits in 9 days, as someone with a 0 coding background," showing traffic and the Cloudflare bill. The author says it was accidental, but these moves are professional.

Where users come from

TimelineWhat happenedRepo stars
11-01Repo created—
11-02Same-title launch on V2EX and NodeSeek. The author's TG channel posted a bilingual card that night, and Chinese Telegram news channels such as LetITFly News reposted it. linux.do and NodeSeek users posted screenshots on their own.—
11-03The forum buzz continued.131
11-04Chinese X user @sosuke997 posted "This test site is pretty interesting... it even shows you how the site sees you," with 109k views and 359 likes. In the replies someone asked whether Grok was safe to use, and Grok answered "open source, MIT, basically no big risk."223 (+92, single-day high)
11-05The author posted a retrospective on X, V2EX and NodeSeek: "9 days, 2M visits." Chinese blogs started reposting it.251
11-06~08The author posted on Japanese Qiita, Korean SIR and English Lemmy. Response was small.268 → 298
11-09~10Included in X's daily "good software" roundup. Portuguese and Russian YouTube Shorts "Top GitHub Projects" lists mentioned it.377 → 425

The site's anonymous analytics match this route (author's stats page, 11-10): over the past 7 days, 184k checks. 62% had system timezone set to China Standard Time, and 65% had browser language Chinese (China). Browsers were 20% X's in-app browser and 5% WeChat's in-app browser. Users are basically Chinese netizens behind proxies, and a large share came in from X posts and WeChat groups. I could not find which group or WeChat Official Account accounted for the 5% from WeChat.

No sign of paid promotion, Hacker News, Product Hunt or GitHub Trending. It barely spread in English-speaking circles.

2M visits, only 400 stars: why?

The 2M visits are self-reported by the author (his own stats show about 1.92M between Nov 1 and 5, with an hourly peak of 50k), and they can't be independently verified. Even discounted by half, every few thousand visits brought in one star.

The logic is simple: people came to check whether their proxy leaked, not to read code. For them the repo has one use: proving the site is open source and doesn't steal data. Some people questioned it in the comments, others asked Grok whether it was safe, and the answer was "open source, MIT." Stars were the trust that came along for the ride. The author opened over a dozen repos for the dozen sub-sites, with READMEs linking to each other, but most of the other repos have only 0 to 54 stars. The cross-links drove traffic and search to the tool site, but did almost nothing for stars.

So don't treat it as a model for "getting stars with an open-source project." A web tool went viral, and the repo rose a little with it.

Who competes with it, and how did they do?

This is a decade-old category. BrowserLeaks ranks around 16k globally, whoer.net around 20k, and BrowserScan around 27k. The Chinese author jason5ng32's open-source IP toolkit MyIP has 12k stars and also checks WebRTC and DNS leaks, but has no portrait and no cross-signal reconciliation. The English siinghd/yourinfo follows the "what the site knows about you" profile route and stops at 562 stars. who.1ms.dog didn't fill an empty slot. It reframed an old category with a new angle and new scenario.

Copycats have appeared and none took off (GitHub, 11-09): yachengzhou/claude-env-check, created 11-08, isolates "Claude/ChatGPT account environment consistency checks plus a one-click fix script" and has 1 stars. xyykmkefke/neon-glass, also created 11-08, covers "what a site can know about you before your first click" and has 0 stars. Most of the 58 forks are people deploying their own copy to use.

If it were you

C. Plenty to learn here, but copying it won't get you stars. This project's stars came from 2M page views riding on the Claude account-ban wave, plus a marketer who writes hook-style titles. The Chinese agent-proxy niche is already taken by it, and the two follow-up repos 11-08 are stuck at 0 to 1 stars.

Good fit for: developers with an existing consumer "check and diagnose" tool (speed, IP, account environment, device health, privacy) who want it screenshotted and shared. Copy the form, not the topic: turn a pile of readings into a personal, human-sounding report, give grades and fixes, make a share card with no private data, and add a "Test me" button.

Bad fit for: anyone planning another Chinese fingerprint or proxy checker (this one holds the spot, and users won't test two of them). Anyone hoping to collect GitHub stars this way (users come to check themselves, not to read code, so star conversion is very low). Anyone without a posting channel or familiarity with Chinese VPS communities.

How to do it: if you still want to play in this space, pick what it doesn't cover. First, desktop and CLI. A V2EX user asked, "Claude Code and similar tools always fetch system info." A web page can't see what the client sees, so an open-source command-line environment checker seems more likely to earn stars from programmers (inference). Second, one-click fixes. Two commenters raised it, and the author admitted it's "fairly weak" here. Think of a Clash rule file you can import directly, or a browser extension that handles timezone, language and WebRTC in one step. (Someone also asked, "Is there a plugin that fixes everything?")

How to get users: copy its launch approach. Use a personal experience as the title. End the post with a question anyone can answer ("How many ways do you leak?"). Post the same day to V2EX, NodeSeek and linux.do, then to your own Telegram channel. On days three and four, write a retrospective with real numbers as a second wave.

When to move: ban-wave interest moves on a weekly cycle. To ride it, ship within November. After December, the discussion fades and "environment checking" is left to longtime users.

Signs you've misread it: if an account-environment-only repo like claude-env-check passes 300 stars within a month, the niche can earn stars on its own and not just through who.1ms.dog. Or if who.1ms.dog's repo keeps adding dozens of stars every week after the ban wave fades, then stars aren't just a side effect of traffic, and this C grade is too low.

GitHub repo ↗

Research notes: competitor table, acquisition evidence, full research report
Competitors
NameWhereLaunchedScaleHow it differs from this one
BrowserLeaksWeb2010 earlyTranco browserleaks.net #15,882 / .com #41,648Raw data listed item by item, with no portrait, grades or fixes
whoer.netWeb2010 eraTranco #19,805Shows a disguise percentage and sells a VPN. Narrow detection, doesn't cover split routing, and doesn't read JA4 / HTTP2 / QUIC
BrowserScanWebAbout 2023Tranco #27,090 (30 days ago #21,992)Scores fingerprint realism, aimed at cross-border e-commerce and anti-detect browsers, with commercial referral links
ping0.cc / ippure.com / scamalyticsWebAlready existedping0 #40,888 (30 days ago #66,845), ippure #158,775, scamalytics #57,894Checks only IP cleanliness, not consistency of the browser environment
Pixelscan / ipheyWebAbout 2020Tranco #292,515 / #221,502English-language, commercial anti-detect browser circle tool
EFF Cover Your Tracks / AmIUnique / CreepJSOpen source2010 / 2014 / 2020amiunique.org #239,630;creepjs 2,546 starOnly answers "Am I unique?", not "Do I look like a proxy?", and gives no fixes
jason5ng32/MyIP(ipcheck.ing)Open source2023-1112,335 stars, +46 in the last 7 days; Tranco #345,590A Chinese author's collection of IP tools, with no personified portrait and no reconciliation of multiple leak signals
siinghd/yourinfoOpen source2025-12562 stars, growth has stoppedSame "what the site knows about you" profile plus a Grok AI profile, English-language, not aimed at proxy or account-ban scenarios
whatdotheyknow.app / privacy.net analyzer / todetect.cnWeb2018–2026SmallGeneric page, weak on detection depth and on Chinese proxy scenarios
Where users come from
  • Verified NodeSeek / V2EX launch (hook title plus personal story): 2026-11-02 Same post, "I run a Japanese node, but the site spots at a glance that I'm Chinese... who sold me out?"; on V2EX, t/1253711 got about 23 hours of 6,591 views and 60+ replies; on NodeSeek, post-1025580 (later 404)
  • Verified Users on the forums sharing screenshots: at least 6 organic NodeSeek posts ("My Claude should be invincible now" 4 pages, "Claude environment showcase", and others); on linux.do, 11-02 "I've become a super hybrid" and 11-04 "How the site sees you"
  • Verified Telegram channels: the author's own channel "1ms.dog / Fast Dog", 11-02 21:00 reposted by LetITFly News; dbzdk, the NodeSeek official channel, 折腾啥 and 福利吧 also reposted it
  • Verified Chinese X organic post: @sosuke's post at 11-04 got 109k views and 359 likes, lining up with +92 stars on 11-04 that day. On-site stats show 20% of traffic came from X's built-in browser. @vansvoler, @chasen_liao and others also recommended it, 997
  • Verified Author's traffic recap (second wave of content marketing): 11-05 long post on X @xjp, V2EX t/1254715, and NodeSeek cross-posted "9 days, 2M visits"
  • Inferred Daily recommendation accounts / GitHub chart short videos / blog reposts: 11-09 X @chuanliang "Daily Good Software"; Portuguese and Russian YouTube Shorts on the Top GitHub charts; 91biu, ZhuLink, ifeed and others; the timing lines up with the +11-08 on 09~79, but the specific contribution of each is unverified
  • Verified Author seeded multilingual communities: Qiita (11-06), Korea's SIR (11-07), Lemmy programming.dev and lemmy.world (11-08), HackerNoon; little response
  • Inferred WeChat groups / Official Accounts: site stats show the WeChat built-in browser at 5%, but no specific source was found
  • Inferred Ads / HN / Product Hunt / trending / Xiaohongshu: not found in the hackernews dataset; the repo was under 60 days old so it couldn't enter trending; no ads or influencer posts found

Why yt1988/who-are-you ("Who are you" / who.1ms.dog) showed up in results

Verdict

This is the open-source repo for a browser fingerprint and proxy-leak self-check page (who.1ms.dog). It took off for two main reasons, plus some luck:

  1. Timing: from 2026-10 to early November, Claude was cracking down hardest on KYC and account bans for Chinese users, and "is my setup clean?" became a shared anxiety across Chinese VPS users, proxy users and the AI crowd;
  2. Product format: instead of listing raw parameters as before, it writes results as a personified portrait, something like "you may be Chinese, currently in Tokyo... your DNS is in China", then gives a disguise grade and a fix for each item. Users can screenshot it to post or meme, and it can also be used to discuss bans;
  3. Distribution: the author has a marketing and media background. He published the same hook title on NodeSeek and V2EX on the same day, also posted to his own Telegram channel, and was then picked up organically by linux.do, big TG channels, and a Chinese X post with 109k views.

GitHub stars were only a by-product of this web traffic. The author says there were about 2M visits in 9 days, yet the repo has only 425 stars. The card's claim of "20+ repo README cross-links driving stars" isn't supported by the data: sibling repos mostly have only 0–54 stars, and nearly all the stars sit in this one repo.

1. What it is

  • The site who.1ms.dog (10 languages). On load it runs 124 "sniffers" with no permission prompt. What it reads includes: server-side IP, ASN, Cloudflare request.cf, TLS/JA4, TCP SYN, HTTP/2, QUIC fingerprints (via a self-hosted probe.1ms.dog); browser-side timezone, language, WebGL/Canvas/audio/font fingerprints, extension traces, Client Hints, and lie detection; and on the exit side, what the domestic sites, foreign sites, chatgpt.com and claude.ai each see as your exit.
  • It assembles everything into a paragraph titled "How the site sees you". On the right are 14 "disguise checks" graded A–F, covering things like timezone vs. IP mismatch, DNS leaks, split routing, WebRTC leaks, datacenter IPs, UA vs. GPU mismatch, anti-detect browser traces, and AI agent and automation traces. Each item comes with a "how to fix", such as Clash fake-ip/TUN, disabling WebRTC, or changing the timezone.
  • There are also share cards and share links (no IP included; the city is only included if you tick it), a "compare with a friend" feature, and an anonymous stats page at /stats.
  • Runs on Cloudflare Workers under the MIT license, self-deployable with npx wrangler deploy. 9 commits in 154 days (GitHub API).
  • It is one tool in the author's 1ms.dog tool-site lineup. The same lineup includes a dozen-plus subsites: speed test, exit check, IP quality, DNS, IPv6, streaming/AI unlock, CDN radar, cloud cost and more, each with its own repo.

2. Data

  • Stars: 11-03 131 → 11-04 223 (+92) → 11-05 251 → 11-06 268 → 11-07 280 → 11-08 298 → 11-09 377 (+79) → 11-10 425 (+48); forks 58. Source: our GitHub repo_day table and data cards. Did not enter GitHub trending (repo under 60 days old).
  • Author-reported traffic: about 1.92M visits over 1–5 November, a single-hour peak of 50k, and 4.6M Worker requests. Source: the author's recap posted on X (@xjp), V2026-11-05EX t/2, and NodeSeek, titled "1254715, a non-coder who built a tool site with 0M visits in 9 days?" (approximate translation of a 2M-visit recap). Not independently verified.
  • Anonymous on-site count (who.1ms.dog/zh-cn/stats, read 2026-11-10): 184,246 detections over the past 7 days. Third-party blogs cite 135k to 138k around 11-05. User profile:
    • System timezone: China Standard Time 62%, Taipei 6%, Singapore 6%, North American Pacific 5%, Japan 4%; browser language: Chinese (China) 65%, US English 16%, Traditional Chinese (Taiwan) 5%. The core base is Chinese users running proxies.
    • Browsers: Chrome 156 at 22%, X built-in browser at 20%, Safari 9%, Edge 7%, WeChat built-in browser 5%. A sizable share of traffic clearly came from clicks on X posts and WeChat.
    • The most common giveaways: split routing 52%, round-trip latency 46%, connection fingerprint 45%, language vs. IP mismatch 37%, timezone vs. IP mismatch 36%, datacenter IP 28%, DNS leak 22%.
  • Domain ranking: who.1ms.dog is not yet in the Tranco top 1M (our traffic dataset, as of 2026-11-09; Tranco is a 30-day rolling list, so it lags by 1–2 weeks anyway).
  • Comparison: as of 11-10, the author's other repos had 1ms.dog at 50 stars, me2cloud at 54, Multi-IP-Routing-Conf at 27, connection-checkup at 18, ip-quality-test at 11, and most others at 0–9 (GitHub API). The main 1ms.dog repo went from 11-03 to 11-10 only, rising from 25 to 50 (github dataset).

III. Competitors

ProductFormLaunchedScaleHow it differs
BrowserLeaksWeb2010 earlyTranco browserleaks.net #15,882, .com #41,648Lists raw data item by item, engineer-oriented, with no portraits, grades or fixes
whoer.netWeb2010 eraTranco #19,805Also has a "disguise percentage" score and sells a VPN; narrower detection, doesn't read JA4/HTTP2/QUIC, doesn't cover split routing
BrowserScan(browserscan.net)WebAround 2023Tranco #27,090 (#21,992 30 days ago, falling)Scores fingerprint realism, aimed at cross-border e-commerce and anti-detect browser users, with commercial referral links
ping0.cc / ippure.com / scamalyticsWebAlready existedping0 #40,888 (#66,845 30 days ago, rising), ippure #158,775, scamalytics #57,894Checks only IP cleanliness and risk, not the browser environment
Pixelscan / ipheyWebAround 2020Tranco #292,515 / #221,502A tool from the anti-detect browser crowd, English-language and commercially oriented
EFF Cover Your Tracks / AmIUnique / CreepJSWeb + open source2010 / 2014 / 2020eff.org #1,686;amiunique.org #239,630;creepjs 2,546 starAcademic and privacy-advocacy oriented, only answers "are you unique?", not "do you look like a proxy?"
jason5ng32/MyIP(ipcheck.ing)Open source + web2023-1112,335 stars, +46 in the last 7 days; ipcheck.ing #345,590The closest Chinese-author open-source IP toolbox: IP, WebRTC/DNS leaks, fingerprinting, speed test; leans toward a tool collection, with no personified portrait or leak-by-leak reconciliation
siinghd/yourinfoOpen source2025-12562 stars (growth has stalled)Takes the same "what the site knows about you" portrait route and uses Grok to generate an AI portrait; English-language, not aimed at the proxy and account-ban scenario
whatdotheyknow.app / privacy.net analyzer / todetect.cnWeb2018–2026SmallAll are generic-style pages on the same concept, with less detection depth and weaker Chinese proxy coverage than this one
Anti-detect browsers (AdsPower, VirtualBrowser, etc.)Desktop software—VirtualBrowser 3,262 stars, etc.These are solutions, while this is a diagnostic tool; they are upstream and downstream of each other and send traffic to each other

Bottom line: "browser fingerprint / leak detection" is a decade-plus old category, and the top sites' Tranco ranks sit between 15k and 40k. It isn't filling an empty niche; it's a new angle and scenario within a mature category.

Four. Where users came from (by time)

  1. 11-01 build. The author "Banana Pi Yang" (X @xjp, GitHub yt1988, NodeSeek uid 8317), self-described former marketer and former media worker, "0 coding background", doing vibe coding with a Claude subscription.
  2. 11-02 morning, first launch in Chinese VPS communities, same copy on the same day:
    • NodeSeek "Tech" section post post-1025580, titled "I run a Japanese node, but the site spots at a glance that I'm Chinese. So I wrote a tool to see who sold me out" (Exa cache; this post later 404, and a top reply said "if it's promotion, just post it as promotion");
    • V2EX t/1253711 with the same title, about 23 hours and 6,591 views, 60+ replies, including someone saying "the Claude ban is making everyone pay attention to browser fingerprinting".
    • The post ends with an engagement hook: "Reply and tell us how many spots your proxy could expose?"
  3. 11-02 same-day organic reposts:
    • At least 6 user-made posts on NodeSeek: "Referencing the boss's project... my Claude should be invincible now" (4 pages), "a site that tests your profile", "come see how the site sees you", "let's check your IP", "Claude environment showcase", "muse shows login region as Guangdong", and others;
    • linux.do 11-02 16:03 "Found a site that tests your profile. I've become a super hybrid", and 11-04 "How the site sees you" (passed around in a group by the DigVPS site owner);
    • The author's own TG channel "1ms.dog / Fast Dog" posted a bilingual card, 11-02 21:00 reposted by LetITFly News; dbzdk, the NodeSeek official channel, 折腾啥 and 福利吧 and other Chinese TG news channels also reposted it.
  4. 11-04, big Chinese X post: @sosuke997 "This test site is pretty interesting... you can even see how the site sees you 😂 who.1ms.dog/zh-cn/", 109k views, 359 likes, 21 reposts; in the replies someone @grok asked whether it's safe, and Grok answered "open source, MIT, basically no major risk". This matches the +92 star peak on 11-04 and explains why the X built-in browser shows 20% in the stats. Other X accounts also spread it, such as @vansvoler (US card and overseas account circles) posting "a big shot on L-site made a fingerprinting site", and @chasen_liao recommending ipqs.1ms.dog as a self-check tool in an Antigravity tutorial.
  5. 11-05, the author posts a postmortem: long X post (@xjp), V2EX t/1254715 (about 1,018 views), synced to NodeSeek. The post shows 1.92M visits and the CF bill, which counts as a second round of content marketing.
  6. From 11-05, Chinese blogs and aggregators repost it: 91biu "差生工具多", ZhuLink 竹林, ifeed, kanmeiba, feedbbs mirror; VicroCode 11-02 also wrote a teardown tutorial, "Rebuilding who.1ms.dog".
  7. 11-06~08, the author seeds multilingual communities: Japanese Qiita (11-06, promoting the 1ms.dog main site), Korean SIR open-source board (11-07), English Lemmy programming.dev webdev/privacy and lemmy.world privacy (11-08, author self-identified), HackerNoon author page. English-side response is small. Our hackernews database has no 1ms.dog post.
  8. 11-08~09 second star peak (+79): three things line up. First, X @chuanliang's "Daily Good Soft Recommendation 2026-11-09" featured Who Are You and linked the repo. Second, Portuguese and Russian YouTube Shorts "Top GitHub projects" lists (the Russian one reads "+238 ☆"). Third, RSS reposts on Fuli Bar. Which one contributed most can't be determined, since there are no stargazer timestamps (GitHub's stargazers endpoint returns 403/404 for our token).

What I didn't find: paid ads, Product Hunt, HN, GitHub trending, Xiaohongshu or Douyin influencers. Site stats show WeChat's built-in browser at 5%, which means it spread through WeChat groups or Official Accounts too, but I found no specific source.

V. Why it took off

  1. Timing, i.e. platform events: late 2026-10 to early November was when Claude KYC and account bans were most frequent. 36kr's "Spending money to buy back banned accounts" (11-03), Huxiu's "A brief history of Claude bans" (10-28), V2EX "my claude account got banned" (5,335 views), and several NodeSeek threads on "KYC twice a week" all fall in this window. Community anti-ban checklists include DNS leaks, timezone/language/node consistency, and browser fingerprinting as standard items, and this page checks all of them at once, with a specific test of what claude.ai and chatgpt.com see as your exit. Over the same period, ping0.cc's Tranco rank rose from #66,845 to #40,888, so demand for "environment checkers" was growing overall, and this tool caught that same demand.
  2. Personified profiles are naturally shareable: identity mismatches like "you might be Chinese, now in Tokyo", "I've become a super hybrid", "pure-blooded Singaporean" are funny, and forum replies are almost all people posting their results, trading codes, and comparing where they got exposed. The three topic types the author summarized in the postmortem are: privacy shock, identity mismatch, and AI-ban anxiety.
  3. It doesn't just tell you you're exposed, it tells you how to fix it: A–F grades, fix steps by Windows/macOS/iOS/Android, and a target-region comparison for "what you want to appear as" turn it from a glance-and-leave toy into a tool you come back to repeatedly. The page stores locally "last time 8 issues, this time 7", which naturally encourages "change it and test again".
  4. The detection surface is deep: JA4, HTTP/2, QUIC, TCP SYN, round-trip latency comparison, lie detection, AI agent mouse trajectories. Hardcore users can't call it shallow. Open-source under MIT plus a "we don't store IPs" privacy note removes the worry of "is this site collecting my data", and even Grok vouched for it: "open source, no major risk".
  5. The copy and distribution are marketer tactics: hook titles ("see who sold me out"), a personal story opening, a closing question for engagement; posted to NodeSeek and V2EX at the same time along with the author's own TG channel; on day 4 a second wave built on the "2M visits" postmortem; then seeded Japanese, Korean and English communities. The author calls it accidental, but the launch playbook was professional.

6. Differentiation, and whether that's why it took off

  • Differentiation: similar tools either list raw data (BrowserLeaks, CreepJS), give only a uniqueness or disguise score (AmIUnique, whoer, BrowserScan), or only check the IP (ping0, ippure). It does three things nobody else combines:
    1. turns hundreds of readings into a plain-language profile;
    2. cross-checks following the logic of "how sites spot a proxy", targeting the problems Chinese users most often hit (split routing, DNS, timezone, WebRTC), and specifically tests the exits Claude and ChatGPT see;
    3. gives an actionable fix for every item, plus a shareable card you can compare with friends.
  • Is that the cause: yes. The profile plus shareable card is why it could spread on forums and X organically; "exposure cross-checks + fixes + Claude exit" is why it stays useful during the ban wave and gets reused. Without the ban wave it would probably be just a fun toy; without the profile format, it would just be another clone of BrowserLeaks.
  • Where the stars come from: the page and README point to the repo, users can audit the code themselves, and stars are a byproduct of that trust. The network of sibling sites and cross-links contributed almost nothing to stars: sibling repos 0–54 stars. The cross-links mainly drive traffic across the whole tool lineup and help SEO. The repo has a separate landing page for each keyword, including /fingerprint/, /dns-leak/, /webrtc-leak/, /proxy-check/, /canvas-fingerprint/, /ai-agent/, /warp-check/, /http3-ech/, and static pages in 10 languages.

7. Visibility and copycats

  • Visibility: spreading within niche circles. Chinese VPS, proxy, and AI circles (NodeSeek, V2EX, linux.do, Chinese TG channels, Chinese X) have widely seen it; the author publicly shared traffic (2M) but there is no revenue, and it was free to begin with; no mainstream media, HN or trending; it barely spread in English-speaking communities.
  • Copycats that have already appeared:
    • yachengzhou/claude-env-check (2026-11-08 created), positioned directly as "local browser check of Claude/ChatGPT account environment consistency, with a one-click fix script", isolating the ban scenario;
    • xyykmkefke/neon-glass (11-08), "what a website can know before you click for the first time";
    • VicroCode's "Rebuilding who.1ms.dog" teardown tutorial (11-02);
    • about 58 forks on GitHub, many of them self-hosted.

8. What's worth borrowing

  • Swap an old category for "plain-language profile + meme-worthy screenshots": any detection or check-up tool (speed test, IP, privacy, account environment, device health) can present results as an identity-flavored narrative with a shareable card, without personal data.
  • Ride platform-event anxiety: when bans, KYC or regional restrictions hit, be first out with a free tool people can use for self-checks right away, and test the target platform directly inside it (claude.ai/chatgpt.com cdn-cgi/trace).
  • Follow the diagnosis with fixes and a "test again" loop, giving the tool a reason to return.
  • Launch playbook: hook titles plus a personal story, posted to several niche communities at once plus your own TG channel; a second wave on day 3–4 built on the traffic postmortem; then seed Qiita, SIR and Lemmy by language.
  • A network of sites for SEO, stars concentrated on one hit: one static landing page per keyword, 10 languages, llms.txt. In addition, Cloudflare Workers with AI-automated operations (issues auto-assigned for fixes) let one person maintain a dozen-plus tools at once.
  • Possible monetization (no one has done it at scale yet): refer traffic to anti-detect browsers, residential IPs, airport proxies, VPSes (this circle's affiliate programs are mature), or sell an "account environment check-up" API. The author currently has no monetization.

IX. Uncertainties

  • The 2M visits and 4.6M Worker requests are self-reported by the author and not independently verified; the 184k detections over 7 days on the site's /stats are the site's own count, not third-party data.
  • The exact source of each of the two star peaks (11-04, and 11-08~09): no stargazer timestamps, so they can only be inferred by matching timing.
  • Why the NodeSeek launch post was 404 (deleted, moved to the promotion section, or deleted by the author) is unclear.
  • The specific source of the WeChat-side spread (5% of traffic), and whether any Official Account reposted it: not found.
  • How much retention and return traffic drops once the ban wave recedes; whether the author will monetize (referring to anti-detect browsers or proxies).

Sources

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