A college graduate's 2000-star project: Address, a real residential address generator. The reputation is real, but the stars are padded
What it is A self-hostable website that generates real residential addresses for 27 countries, with postal codes that match and locations you can find on a map. It's mainly used for registering overseas accounts, opening cards, and buying eSIMs.
Why it took off The author posted an open-source promo on LINUX DO and got 235 likes. Add X posts from Chinese "free-perk" accounts and blog reposts, and it built a reputation in small circles. The surge in early November has no traceable external source and matches star-farming patterns, not organic growth.
CHow you could play it Don't build another full-featured version. Build what it does poorly: a lightweight deployment (split data by country, runs on a Cloudflare Worker or a small machine), launch in LINUX DO's open-source promotion section, and don't farm stars.
- Data (updated daily) · Repo created 2026-07-15 · 2,153 stars · 7 days +440
GitHub · stars Data 2026-10-09 → 2026-11-08 (31 days, 17 observations). Points before 10-28 come from deeply / archive backfill, not every day has one
daimon3332/address is a self-hostable "real residential address" generator covering 27 countries. Four months after open-sourcing, it has over two thousand stars. It does have real users in a small Chinese circle, but its star curve doesn't add up: late October it lost over five hundred stars overnight, then early November it rose back in sync with the same author's other repo on the same day. Anyone trying to earn stars with a similar project can learn how it built reputation in the community. What they shouldn't copy, and can't, is that curve.
A college graduate posted two threads on the same morning
On 2026 Jul 25 at 5:55 (UTC), LINUX DO user daimon1 posted a question: "If GitHub star count goes on a resume, how impressive is it? (college applicant)" His own answer: "500+ should be worth listing." The thread got 723 views.
Less than three hours later, he posted a second thread in the forum's "Resource Collection" section, titled "Stop filling forms with fake streets: self-hosted real open-data addresses for 27 countries + test data generator." Tagged "Open Source Promotion," it ends with: "If the project helps you guys, please kindly give it a star with your rich little hands." So far it has 2,018 views, 235 likes, and 40 replies, which is high for a tool post.
So from day one, the project knew exactly what it was after: not money, but a star count you can put on a resume.
What it is and what people use it for
The README describes it as an "address and synthetic test data generator": roads, administrative regions, coordinates, and postal codes come from official national address registries, OpenStreetMap, OpenAddresses, and other open data. Each record states its precision, and a field is only labeled "residential" when independent evidence of a residence exists. Fields that can't be verified are left blank. It also generates test data like names, jobs, and sandbox cards. 10 display languages are supported. Chinese addresses are precise to the housing estate, and you can check them with one click on Google Maps or Amap. The stack is Astro + Hono + PostgreSQL, runs from a single Docker Compose file, and includes an admin panel, API tokens, and OpenAPI docs. The database was built in Jul 15, and it reached v2.4.1 by Nov 4, with 15 versions in between.
The replies under the post make the real use case plain. Someone said, "Just search for the Chinese embassy address in that country." The author replied: "An embassy isn't a residential address. For opening overseas banks or U cards, you need a residential address to pass more easily." Other replies: "I use it for buying eSIMs and registering Apple IDs"; "A lot of address generators produce addresses where the postal code doesn't match"; "At least there are no ads. Those other sites cram the whole page with them." One third-party deployment tutorial is titled "Breaking the Cross-Border Risk Control and AVS Blind Spot."
In other words, it solves a small problem that overseas registration people hit over and over: finding a real residential address with a matching postal code and city, one you can see on the map.
The author's other open-source projects run along the same line: Outlook bulk registration (OutlookRegister), an Outlook management and token-retrieval tool, a multi-port proxy pool he calls "essential for registration tools" on the forum (easy-proxies), registration scripts for GPT and Grok, and temporary email. Address is the "address" step in that pipeline.
Where the stars came from: the visible part
The exposure you can find is almost entirely in small Chinese circles:
| Timeline | Where | What they said | Results |
|---|---|---|---|
| 07-25 | LINUX DO "Resource Collection" open-source promo thread | The launch, with a request for stars in the post | 2,018 views, 235 likes |
| 07-27 | X Chinese account @NFTCPS (crypto and freebie-focused) | "The pain of filling out forms with fake addresses and fake info... let me recommend an open-source project" | Replies included "Getting stuck on the address field is so annoying every time"; view count unknown |
| 08-21 | LINUX DO "Develop" section | V2, a full rewrite with map verification added | 542 views, 42 likes |
| 08-27 | NodeSeek (VPS community) | The same post moved over, titled "Guaranteed all real residential addresses!!!" | Engagement unknown |
| 8–September | Chinese blogs and resource sites | A few intros and deployment tutorials | Brought in some organic search traffic |
| 10–November | A batch of content farms | Auto-generated "complete explainers" that scrape GitHub | Didn't bring in many people |
Added up, the publicly visible views total only a few thousand. But working backward from the star timestamps on trendshift.io, it already had about 1,500 stars in early September, which means most stars were accumulated in 7, in August. A few thousand views turning into over a thousand stars is an unusually high conversion rate (inference).
The curve that doesn't add up
Based on GitHub's daily star records:
- 10-13:2,215 stars.
- 10-28:1,704 stars. Over these two weeks trendshift recorded only about 50 new additions, so roughly 560 stars actually disappeared. In late October GitHub ran a large-scale account cleanup. Among nearly 4,900 repos with over 200 stars and before-and-after data, only 32 dropped more than 15%, and many were obvious star-farming repos wiped down to near 0 overnight. This author had two repos on that list: Address dropped 23%, and OutlookRegister fell from 1,268 to 746, down 41%. His other repos barely moved.
- From 11-04 to 11-05: one day +190 (trendshift dated the jump to 11-04, +222), while forks only rose by 2. The same day OutlookRegister gained +164, and none of its forks grew. 11-06 Two repos rose together again, +93 and +76. Of the 32 repos that were cleaned up, only 3 had single-day growth of over 5% in 11-05, and two of those were his.
- Around 11-04, no new recommendation posts appeared on X, LINUX DO, or blogs. That day it only shipped a minor version. The X post speaking up for it (@knowledgefxg "found a useful open-source project, Address") was published on 11-06, after the surge.
- From 10-28 to 11-06, stars rose about 300, but forks only went from 321 to 331.
GitHub no longer lets outsiders inspect star accounts one by one, so it's impossible to confirm who starred it. Stars removed in cleanups usually come from banned accounts, and this community has long run large batches of registered alt accounts, so the author may not have done it intentionally. But "two of the same author's repos cleaned up together, rebounding together, forks flat, no traceable external source" matches the pattern of star-farming or mutual-star groups (inferred, moderately high confidence). The takeaway is simple: its reputation is real, but its star count shouldn't be read as a sign of demand.
Who it's competing with
Searching GitHub for "address generator / 地址生成" ranks it first by stars. Here are other approaches to the same need (star counts as of early November):
| Project | Scale | Compare with Address |
|---|---|---|
| GuooGaii/ip-geoaddress-generator | 580 stars / 662 forks | Generates real addresses from IP, mainly US; runs on Cloudflare Pages easily. Light, narrow coverage |
| chatgptuk/Real-US-Address-Generator | 364 stars / 394 forks | US only, runs on a CF Worker |
| hotbob011/real-random-taxfree-address | 176 stars | Focused on US tax-free state addresses, very targeted |
| ALioooon/RealAddressGenerator | 42 stars | Claims 44 countries, but addresses are assembled from rules |
| Web tools like meiguodizhi, fakexy, fakenamegenerator | Rank high in search | Lots of ads; addresses are often made up, zip codes don't match |
| Faker / faker-js | 19k / 15k stars | Generate fake addresses that don't exist, meant for testing. Not the same thing |
Look at the first two: forks outnumber stars, so people are really deploying it themselves. Address's weak point is also deployment: the author admitted the database shrank from an estimated 40G to 9G, "you really need a big machine"; one reply said "a small box probably won't work," and someone froze a 2C2G machine.
Three things worth learning
- Find one small thing a community keeps having to work around by hand, and make it genuinely good. Here it's "one real residential address." The pitch isn't the count "44 countries," it's "every entry can be opened on a map and shown as a real house." Verifiable real beats large numbers.
- Launch in the right community. LINUX DO's "open source promotion" section is friendly to open source projects. One tool post got over two hundred likes; August's V2 push got another round, and in September blogs and resource sites reposted it.
- Build a set that feeds each other. Registration, email, proxies, addresses: they target the same crowd, so a project's readers click over to the next one.
If it were you
C: you can learn from it, but copying it isn't worth the effort. The real reputation comes from a very narrow Chinese circle, and Address already holds that spot. It's unclear how many of its 2,000 stars are real, so using it as a benchmark for "how many stars this kind of project can get" will overestimate.
- Who it's for: developers already hanging out on LINUX DO, NodeSeek and similar communities who understand overseas signup and risk-control pain, and want a project with a few hundred stars for a resume or to grow followers.
- Who it's not for: anyone wanting a general project that earns stars on English GitHub (that crowd has no interest in "risk-control-passing addresses"; test data already has Faker, with tens of thousands of stars). Also anyone who cares about a clean reputation and resume (this category sits right next to registration bots and bulk alt accounts, which may not help on a resume).
- How to build it: don't build another "all 27 countries" version head-on. Address's failing is that deployment is too heavy. The most worthwhile move is a lightweight version: split the data into per-country small packages that fit in a Cloudflare Worker or a single SQLite file, runnable on a 1G-RAM small machine or even pure frontend; or cover just one or two countries at street-number precision with map verification. The 662 forks of ip-geoaddress-generator show people want "light and self-hostable."
- How to get users: launch in LINUX DO's "open source promotion" section, tag it per community rules and disclose the AI-generated parts; two weeks later post a major-version update; then go to NodeSeek and Chinese sharing accounts on X. Don't buy stars. This cleanup showed that inflated stars get wiped in one go, and the same author's other repos get watched too.
- When to do it, before what: no rush, the demand has always been there. But if Address builds its own lightweight deployment in V3, this gap is gone.
- What would mean I read it wrong: (1) Address organically climbs past 3,000 stars in a month with no new external exposure, forks rising with it. That would mean its traffic is larger than what's visible from outside, and this category deserves more serious work. (2) Conversely, GitHub runs another cleanup and it drops again. Then even the "a few thousand views for a thousand-plus stars" path doesn't hold, and a small-circle tool post probably only gets a few hundred stars.
Research notes: competitor table, acquisition evidence, full research report
- Competitors
Name Where Launched Scale How it differs from this one Web generators like meiguodizhi / fakexy / fakenamegenerator Web Years Rank high in Chinese search; traffic unverified They're full of ads, addresses are assembled, and zip codes don't match. Address's addresses are real and traceable, and it can be self-hosted with an API. GuooGaii/ip-geoaddress-generator Open source 2024-09 580 stars / 662 forks IP-based generation, mainly US, very light to deploy. Address covers 27 countries with residential evidence, but deployment is heavy. chatgptuk/Real-US-Address-Generator Open source 2024-08 364 stars / 394 forks US only, runs on a CF Worker. Address is multi-country, multilingual, with data sync. hotbob011/real-random-taxfree-address(mockaddress.com) Open source 2026-03 176 stars Focused on US tax-free state addresses. Address covers more ground. ALioooon/RealAddressGenerator(addressgen.top) Open source 2025-07 42 stars Claims 44 countries but generates addresses by rule. Address's every entry has a real data source. Faker / faker-js / randomuser.me Open source Years 19.5k / 15.5k / 1.5k stars The addresses they generate don't exist. Address's addresses are real. OpenAddresses Open source 2014 3.3k stars It's the upstream raw data, with no generator or UI. - Where users come from
- Verified LINUX DO open-source promo posts (self-posted by the author): topic 07-25 with 2651060:2,018 views and 235 likes, asking for stars in the post; topic 08-21 with 2788886:542 views and 42 likes
- Verified Chinese X accounts promoting it: @NFTCPS 07-27 tweeted "here's an open-source project I recommend: Address address generator"; @knowledgefxg 11-06 also posted
- Verified NodeSeek and Chinese blog/resource-site reposts: nodeseek post-896565(08-27), jamecling, lijianfei, itimo, 宅叔小站, DC 娱乐网, etc. (Exa search)
- Verified Same author's registration-tool portfolio cross-linking: OutlookRegister, easy-proxies ("essential for registration bots"), gpt / grok-register, etc., aimed at the same audience
- Inferred Possible star inflation or mutual starring: during the late-October GitHub cleanup, address lost about 510 stars and OutlookRegister lost about 520; 11-04 both repos surged on the same day with forks barely moving, and no external source was found. The author 07-25 asked "how many stars are enough to put on a resume?"
Address (daimon3332/address): why it took off
Channel: GitHub · Research date: 2026-11-06 · Data card: gh_daimon3332/address
Bottom line
Address is a self-hostable "real residential address" generator covering 27 countries. It's billed as for testing, but in the Chinese community people use it to register overseas accounts: opening overseas bank cards and U cards, buying eSIMs, signing up for Apple ID, and other cases that require an address. The author daimon (called daimon1 on LINUX DO, a level-3 user) is a campus recruit whose open-source projects all revolve around the "registration bot" ecosystem: bulk Outlook registration, proxy pools for registration tools, Grok and GPT registration. The project's earliest traffic came from two "open source promo" posts he made on LINUX DO, plus a Chinese X account with a big following and a round of reposts from blogs and content farms. But its star curve isn't clean. When GitHub mass-cleaned a batch of accounts in late October, it lost about 510 stars (about 23%), and the same author's OutlookRegister lost about 520 (about 41%). The +222 stars from 11-04 brought only 2 forks, and it surged on the same day as OutlookRegister; no promoter shows up in public channels. The data card suggests this surge probably wasn't organic. The product itself has real demand and real users (people deploy it, file issues, write tutorials), but star counts shouldn't be read as a heat signal.
1. Data
- Repo created 2026-07-15, MIT license, JavaScript (Astro + Hono + PostgreSQL), one-command deployment with Docker Compose. From v2.0.0 (08-03, full rewrite) to v2.4.1 (11-04), 15 releases total (GitHub releases API).
- Star curve (github repo_day / repo_snap):
- 10-09:2152 → 10-13:2215 (deeply source, +8 to +25 per day)
- 10-28:1704 (census source). trendshift counts by starred_at, 10-13 to 10-27 added about 50, so over this period about 560 stars actually disappeared
- 11-04:1737 → 11-05:1927 (+190) → 11-06:2020; the GitHub API read 2108 at 11-06 14:30 UTC
- fork: 10-28 has 321,11-06 and 331; stars rose about 300, but forks only added 10.
- trendshift.io (counted by star timestamp, UTC): in September, 0 to 6 per day; from 10-06 to 10-12 a small wave of 18 to 36 per day; 11-04 +222 (forks +2), 11-05 +104 (forks +8), 11-06 +44. 9-07 to 11-06 totals +630. Also, trendshift's records show it never made the top 25 on any single day.
- Working backward: at 9-07 it had only about 1500 stars, meaning most of the stars were accumulated from 7 to August. But the public exposure findable for that period totals only a few thousand views (see section 2).
- Today GitHub's stargazers endpoint returns 404 for every repo, so I can't check which accounts starred it (unverified).
1.1 Late-October star drops
I pulled the 4,876 repos in the github table that were observed between 10-13 and 10-27 with ≥200 stars, and compared them with the 10-28 census (github table, SQL check):
- Only 32 repos dropped more than 15%. Many were obviously inflated, cleaned back to nearly 0 overnight, such as RubinLabs26/ISpotify going from 1896 to 6, and wangmingxuan666/Tiersense going from 920 to 13.
- Two repos from the same author are on this list: address went from 2215 to 1704 (−23%), and OutlookRegister went from 1268 to 746 (−41%). His other repos, OutlookManage, Outlook-Oauth-GetToken and easy-proxies, barely moved.
- 11-05, of these 32 repos only 3 gained more than 5% in a single day, and two of them are this author's: address +190, OutlookRegister +164, while an OutlookRegister fork did not gain. 11-06 the two repos also rose together +93 and +76.
- Stars that GitHub cleans out usually came from banned accounts. Two repos rising on the same day, forks barely moving, and no matching posts found online fits the pattern of star farming or mutual-starring groups. But I can't see which accounts gave the stars, so this can't be confirmed. Another possibility is that a batch of wrongly banned accounts was restored. But trendshift records by the new star timestamp to 11-04, which doesn't look like restored old stars, so that possibility is less likely.
2. Where users come from
| Timeline | Channel | Item | Results / evidence |
|---|---|---|---|
| 2026-07-25 08:38 | LINUX DO "resource roundup" open-source promo post 2651060 | "Stop filling forms with fake streets: self-hosted real open-data addresses for 27 countries + a test-data generator." The post openly asks "bros, give it a star with your little hands" | 2,018 views, 235 likes, 40 replies (post JSON) |
| 2026-07-25 05:55 (same morning) | LINUX DO post 2650329 | The author asked: "For GitHub stars, if I put them on my resume, how many look good? (campus recruit)" His own conclusion was "500 or more should be fine to list" | 723 views. This shows he's chasing stars to put on his resume |
| 2026-07-27 | X @NFTCPS (Bluebird Club, Chinese crypto and deals account) | "Ugh, the pain of filling out a form and having to make up a fake address and fake details 😅 Here's an open-source project I'd recommend..." with a live demo link | A reply said: "Getting stuck on the address field every time is really annoying." It got no views. |
| 2026-08-21 | LINUX DO "Develop" thread 2788886 | V2 rewritten release: "27 residential address generator, with location checks via Google Maps / Amap" | 542 views, 42 likes |
| 2026-08-27 | NodeSeek post-896565 | The same post moved to the VPS crowd, with the title "Guaranteed all real residential addresses!!!" | Engagement count not retrieved |
| 8 to September | Chinese blogs and resource sites | jamecling's software blog (09-01), lijianfei.com (09-14, titled "Breaking Cross-Border Risk Control and the AVS Blind Spot"), ITIMO, Zhai Shu's site, DC Entertainment Network, macin.top deployment tutorial, vvars | Exa search |
| 10 to November | Content farms | Auto-generated "complete breakdown" posts on sites like mhpq.cn, xxqk.cn, lpnh.cn, xdnf.cn, hotelren.com, which keep appearing (11-06) | SEO posts auto-scraped from GitHub. They don't bring in much traffic. |
| 2026-11-06 12:16 UTC | X @knowledgefxg (Knowledge Sharing Official) | "Found a useful open-source project, Address..." | Posted after the 11-04 surge, so it wasn't what caused that one |
- How users actually use it, per the LINUX DO replies: "For opening foreign bank accounts or U cards, a residential address still gets through more easily" (the author); "I use it for buying eSIMs and registering Apple IDs" (wty); "A lot of address generators give addresses where the zip code and address don't match" (kant); "At least there are no ads, unlike those sites that cram the whole page with them" (zzqq9921). lijianfei.com's title just says "Breaking Cross-Border Risk Control and the AVS Blind Spot." The README says form testing, but the real need is getting past risk control: a residential address, with zip code and city that match, that can be found on a map.
- The author's whole set of open-source projects is in the same ecosystem (GitHub profile): OutlookRegister (bulk Outlook registration, 1022 stars), OutlookManage, Outlook-Oauth-GetToken, easy-proxies (titled "Multi-port proxy pool, a must-have for registration bots" on LINUX DO), gpt-free-register, grok-register, tempmail. His LINUX DO post history also includes topics like "long-term PayPal-linked channels," "GPT Team 48-month discount," "Gemini student verification," and "paid request for Outlook K12 Workspace." Address is the address step in this "bulk overseas account registration" pipeline.
- Traces of real users: issues #16, #17, #18 (11-02 to 11-04) are all deployment errors and map data out of sync (Brave search results). On LINUX DO someone deployed it on a 2C2G small machine and froze it (Si1ence, 08-24). There are also several third-party deployment tutorials.
- 11-04 this surge of +222 has no external source: it isn't in the local X corpus (x-radar), and Jina / Exa / Brave turned up no new posts from 11-03 to 11-04. That day only a minor version v2.4.1 shipped.
3. Competitors
| Product | Where | Launched | Scale | How it differs from Address |
|---|---|---|---|---|
| Web generators like meiguodizhi.com, fakexy, fakenamegenerator | Web | Years | When Chinese users search "US address generator," these come up first. Traffic not verified. | Lots of ads, and addresses are often assembled, so zip code and city don't match. Address uses real residential addresses with sources, and you can self-host it with an API. |
| GuooGaii/ip-geoaddress-generator | Open source | 2024-09 | 580 stars / 662 forks (github repo) | Generates real addresses from IP, mainly US, runs on Cloudflare Pages, very light. Address covers 27 countries and has residential tags, but needs 4GB+ RAM and 20 to 30GB of disk. |
| chatgptuk/Real-US-Address-Generator | Open source | 2024-08 | 364 stars / 394 forks | US only, runs on a CF Worker. Address is multi-country, multilingual, and has data sync. |
| hotbob011/real-random-taxfree-address(mockaddress.com) | Open source + web | 2026-03 | 176 stars | Focuses on US tax-free states, very targeted. Address has broader coverage. |
| jiangnan1224/AddressGenerator | Open source | 2025-06 | 122 stars | Simpler features |
| ALioooon/RealAddressGenerator(addressgen.top) | Open source + web | 2025-07 | 42 stars | Claims 44 countries, but addresses are generated by rules. Every Address entry has a data source. |
| Similar posts on LINUX DO | Community | 2024 to 2025 | "Real address generator 2.0," "MockAddress," and several other posts | All are web tools. Address's selling points are self-hosting and an open-source database. |
| Faker (Python, 19.5k stars) / faker-js (15.5k stars) / randomuser.me | Open-source library | Years | Big | Generate fake data, and the addresses don't exist. Address's addresses are real. |
| OpenAddresses (3.3k stars) | Open data | 2014 | Big | One of Address's upstream data sources. It provides raw data only, with no generation or UI. |
Searching GitHub for "address generator / 地址生成," Address ranks first by stars (github repo search). Second place is ip-geoaddress-generator, with only about 1/3.
4. Differentiation
- The addresses are real, and their sources can be traced. Data comes from national government address registries, OSM, OpenAddresses, and Google Geocoding. Each entry is labeled with its precision (street-level or house-number-level). Something is only marked residential when independent residential evidence is found. Fields that can't be verified are left empty, not made up. Most alternatives either produce random strings or only cover the US.
- Can be self-hosted with a single Docker Compose file, with an admin panel, API tokens, and OpenAPI docs. 10 display languages. Chinese addresses are precise to the housing estate. Results can be checked directly on Google Maps or Amap.
- These differences really do address users' pain points (residential addresses, matching zip codes, no ads), and the praise on LINUX DO is for exactly these. But they don't explain 2000 stars on their own. The known public exposure adds up to only a few thousand views. The October cleanup removed about 1/4 stars in one go. The 11-04 surge still has no traceable source. The product built a reputation in a small circle, but star count is a different matter.
5. Verdict
- Real demand: "real residential addresses for getting past risk control" is a must-have in the Chinese deal-hunting and registration-bot circles, and people have been building for it for a long time. Address is the most complete open-source implementation in this category. Its first LINUX DO post got 235 likes, which is high for a tool post.
- The stars are inflated: there was a large GitHub cleanup, the author's two repos rose and fell together, forks didn't rise with them, and on the same day the author shared the project, they asked "how many stars can go on a resume?" The data card says 11-05 just started picking up, and that judgment is most likely a false signal.
6. What others can learn
- In "registration bot / deal" communities like LINUX DO, doing one small thing that people keep having to work around by hand (finding a real residential address), doing it most truthfully, most completely, with no ads, and self-hostable, can earn very high community reputation.
- Make "authenticity" a verifiable selling point: every address can be opened on Google Maps to see the house. That is more convincing than a count-based claim like "44 countries."
- Package it as a toolkit: registration bot, proxy pool, email, address, each feeding traffic to the others.
7. Still unclear
- Where the 222 new stars for 11-04 came from: the stargazers endpoint won't load, so we can't see the accounts. Star farming, mutual-star groups, or some Telegram channel we didn't catch, we can't tell.
- 7 to August: how many of the 1500 stars were real. And whether the batch wiped in October was the author's own inflation, or only stars from banned bulk-registration accounts (this circle has plenty of sock accounts).
- Real usage of the live demo site, address.333186.xyz.
- View count on @NFTCPS's post on X.
Sources
- Data card; github repo tables repo_day / repo_snap (10-09 to 11-06); trendshift.io/repositories/94306 (11-06 read)
- GitHub API: repos, releases, the user's repo list, forks (11-06)
- LINUX DO: topics 2651060, 2788886, 2650329, and /topics/created-by/daimon1.json (read via Jina)
- X: x.com/NFTCPS/status/2081703991382331665, x.com/knowledgefxg/status/2118678943859650638 (read via Jina)
- Exa / Brave searches: nodeseek.com/post-896565-1, jamecling.com/archives/6081, lijianfei.com/post/self-hosted-real-residential-address-generator/, itimo.cc/post/111515, macin.top/posts/5d520128/, etc.