Lost 522 stars, then gained them back within a week: how did an Outlook registration tool get its stars?
What it is: a Python script that batch-registers Outlook/Hotmail mailboxes through the browser and automatically gets the OAuth2 token needed to read mail. It's a secondary development of someone else's project.
Why it took off: AI products need large numbers of mailboxes to register throwaway accounts, and many people in Chinese forums were asking for a registration tool. The author pushed a whole set of companion tools on linux.do all at once. But most of its 1,000 stars look bought: during a cleanup it lost 41%, and afterward it surged in sync with the author's unrelated repos while the companion repos stayed flat.
CHow you could approach this: don't copy the browser-based registration tool. Instead, study lxf746, which takes a pure-protocol route, or build a clean component of the pipeline (proxy pool, account management, token health checks, public success-rate tests), and launch it as a bundle in linux.do's open source promotion section.
- Data (updated daily) · Repo created 2026-07-23 · 1,117 stars · 7 days +365
GitHub · stars Data 2026-10-09 → 2026-11-08 (31 days, 25 observations). Points before 10-28 come from deeply / archive backfill, not every day has one
In 2026 July, a campus recruit built a repo for batch-registering Outlook mailboxes in two days, then asked a forum how many stars he'd need to make his résumé look impressive. Three months later, during GitHub's cleanup of star-boosting accounts, the repo suddenly lost 41% of its stars, and a week later it gained them back on the same days as his other, unrelated repo. Below: what it is, where the real demand lies, which of its stars are real, and if you want real stars in this space, whom to watch and what to learn.
A post: "How much does it impress?"
In 2026 Jul 25, a linux.do user named daimon1 posted a thread titled "GitHub star count: if I put it on my résumé, how impressive is it?" The body had one line: "Split into single repo and all repos (campus recruiting, not experienced hires)."
The replies were blunt. Some said "100k+ I guess, anything less and it looks fake," and others said "star counts are flooded with junk now, so their reference value has dropped sharply." daimon1 replied: "Yeah, that's true... I think if you get to 500 or more stars you can probably still list it." As of 11-06 the thread had 723 views.
Two days earlier, on Jul 23, his GitHub account daimon3332 created OutlookRegister. On Jul 24, he also created a repo named github-star-mointor.
What it actually is
A Python script: it opens multiple browsers (using patchright, a variant of Playwright with automation fingerprints removed) to batch-register @outlook.com / @hotmail.com mailboxes, attempts Microsoft's "press and hold" human verification (fully automatic, semi-automatic, and manual modes), and then uses Microsoft Graph's OAuth2 to get a refresh_token, writing each result to a txt file as one line: email----password----client_id----refresh_token. This format is common in the registration-tool community and among people who sell accounts.
It isn't original. It's a secondary development of LainsNL/OutlookRegister, adding secondary-email binding, batch registration, and a README in Simplified Chinese, Traditional Chinese, and English. The README heavily promotes three companion tools by the same author: Easy Proxies (a multi-port proxy pool), Outlook-Oauth-GetToken (gets tokens for existing accounts), and OutlookManage (a web dashboard for managing accounts). Together they form a pipeline: proxy → register → get token → manage accounts.
It has 8 commits in total, all between 7-23 and 7-27. Apart from the first one, they're all doc edits, community link additions, and icon size tweaks. It hasn't been touched since. The issues are all users reporting the same thing: "can't pass human verification," "verification code won't go through," "device or network settings are blocking you from continuing." The author hasn't replied to any of them (GitHub, 11-06).
Who wants this
The demand is real. Throughout 2026, "registration tools" have been hot in Chinese communities: ChatGPT / Codex, Grok, AWS's Kiro, and various API relay sites all have people batch-registering throwaway accounts to farm free quotas. Each account needs a mailbox, and Outlook / Hotmail are the most common source. On linux.do, posts like "is there an open source Outlook registration tool?" "Blood, sweat and tears just to register an Outlook account" and "Outlook registration tool stuck on human verification" are everywhere, and it's the same on NodeSeek.
The author is embedded in this community. He has forked gpt-free-register and reg-factory, and he also wrote grok-register. From 7-24 to 7-26, he posted several threads in linux.do's "Open Source Promotion" section pushing his companion tools: Easy Proxies, "a must-have multi-port proxy pool for registration tools," 952 views, 62 likes; GetToken, 527 views; OutlookManage, 424 views; and an address generator was the hit, with 2019 views and 235 likes. The README links to linux.do and NodeSeek "community endorsement" pages, which is the norm for that section.
There's also word of mouth. In a thread asking for a Hotmail registration tool, someone replied "find OutlookRegister on GitHub and tweak it," and the second half of the same reply said "the success rate is a bit low."
Searching its name on X returns 0 posts from the past 120 days. No media outlet has written about it; the only hits are auto-indexing sites like DeepWiki.
What the star curve says
Daily GitHub star counts, with it alongside several repos from the same author and the upstream project:
| Date | OutlookRegister | address (same author, address generator) | easy-proxies (same author, the README says to use it alongside) | OutlookManage (same author) | Upstream LainsNL |
|---|---|---|---|---|---|
| 10-13 | 1,284 | 2,215 | – | 127 | – |
| 10-21 | 1,268 | – | – | 136 | – |
| 10-28 | 746(−522) | 1,704(−511) | 202 | 137 | 1,334 |
| 11-04 | 782 | 1,737 | 204 | 142 | 1,344 |
| 11-05 | 946(+164) | 1,927(+190) | 205 | 142 | 1,347 |
| Night 11-06 | 1,113 | 2,132 | 209 | 142 | 1,353 |
Three things to look at together:
- 10-21 to 10-28, GitHub ran a site-wide cleanup of star-boosting accounts. In the same window, many repos lost between 40% and nearly 100% of their stars (some went from 3,213 to 242, others from 1,897 to 6). OutlookRegister lost 522 stars, 41%; the same author's address lost 511. What got cleaned up were the stars from fake accounts.
- The 11-05 rebound doesn't look like real traffic. A registration tool and an address generator have nothing to do with each other, yet each gained 160–190 on the same day. Easy Proxies, which the README says is "required" alongside, gained only 2 over two days, and the account management dashboard gained 0. If people were really drawn in by the registration tool, the companion tools would rise with it. The upstream gained about 3 a day over the same period. Of the cleaned-up repos, very few kept climbing rapidly afterward.
- Forks didn't drop. Forks stayed between 107 and 110 before and after the cleanup, which suggests the forks are mostly real people. Roughly 110 forks per 1,100 stars, a ratio of 0.1. The upstream is 382 forks to 1,353 stars, about 0.28.
Outsiders can't verify the accounts that starred it one by one, so "star boosting" is an inference drawn from the shape of the ups and downs, and it's a strong one. To be conservative: the 746 left after the cleanup is its natural ceiling for stars, though some boosted stars may still remain.
Who it's competing with
| Project | Launched | Scale (11-06) | How it differs |
|---|---|---|---|
| LainsNL/OutlookRegister | 2025-06 | 1,353 stars / 382 forks, 10-31 still being updated | Its upstream, which has been maintained the whole time |
| lxf746/outlook-auto-register | 2026-08 | 1,160 star / 469 fork | No browser; it calls Microsoft's web endpoints directly, which is faster and lighter. It didn't drop during the cleanup and instead rose from 937 to 1,157 |
| V-IOLE-T/hotmail-register-extension | 2026-04 | 325 star | Chrome extension format, aimed at OpenAI registration |
| Older English projects like MatrixTM/OutlookGen | From ¥2022 | 441 stars, etc. | Mostly outdated and don't handle OAuth tokens |
| Small Chinese projects like twfb, rain-fly, Iron-Yzg | 2020–2026 | Dozens to 144 stars | Assorted small scripts for registration plus token retrieval |
| Account sellers | Always present | There are claims that Outlook token accounts sell for a few cents each (unverified) | Just buy finished accounts; someone on the forum said "a few cents each, why bother with a registration bot" |
The one really worth watching is lxf746: launched a month later, about the same stars, forks more than 4x as many, and it didn't drop during the cleanup week, it kept climbing. Same need, same crowd, this is what organic growth looks like. It wins on technical approach: everyone else gets stuck on the human verification, it skips the browser entirely.
If it were you
Verdict: OutlookRegister itself isn't worth copying. Most of its stars didn't come from the product; but the need it serves is real, and someone already got a natural thousand stars off the technique.
Who it's for: Chinese developers who hang out on linux.do / NodeSeek, understand browser automation or packet capture, and want to build open-source reputation; who can live with their project sitting in a gray area and at risk of Microsoft breaking it with a update, or even GitHub taking it down.
Who it's not for: anyone who wants a project they can list on a resume and explain to an interviewer. "Bulk-registering Microsoft burner accounts" is a hard sell in a serious interview. Also not for people without time to keep fighting Microsoft's risk controls, since scripts like this stop working within a month of going unmaintained.
How to build it: don't make the Nth "browser + beat the human check" registration bot. What's worth doing is the parts the competitors haven't done well: publish a reproducible success-rate test (everyone asks about it in the Issues), build the clean pipeline pieces (proxy pool, account management backend, batch token refresh and health checks; the author's own proxy-pool post got 62 likes, more than the bot itself), or bring the same approach to other email providers.
How to get users: copy the author's playbook. Post a whole bundle of interlinked tools in linux.do's "open-source promotion" section, with READMEs linking to each other. Then reply to the "registration bot stuck on human verification" help threads. Don't buy stars: this cleanup dropped 41% of them, and synchronized rises and falls are easy to spot.
When this window closes: as long as AI products keep handing out free credits and people keep bulk-registering, the need exists. If the major AI products tighten free tiers or make phone-number verification mandatory, demand will shrink quickly.
What would mean we misread it: if OutlookRegister keeps gaining dozens of stars a day for the next few weeks with no updates or promotion, and its companion repos and forks rise along with it, there's a real traffic source we're not seeing, and the "star-buying" call above needs revising.
Research notes: competitor table, acquisition evidence, full research report
- Competitors
Name Where Launched Scale How it differs from this one LainsNL/OutlookRegister Open source 2025-06 1353 stars / 382 forks, still being updated Upstream of this repo; this repo adds secondary-email binding, multi-stage login, cookie-first OAuth, batching, trilingual READMEs, and a companion toolchain lxf746/outlook-auto-register Open source 2026-08 1157 stars, steadily rising during the cleanup (937→1157) Pure protocol, no browser; more advanced technical approach, and its star curve looks more organic V-IOLE-T/hotmail-register-extension Chrome 2026-04 325 star Chrome extension form, targeting OpenAI registration and account-tag management Older English projects like MatrixTM/OutlookGen Open source 2022 441 stars, etc. Old English-language projects, mostly outdated and not handling OAuth tokens Wholesale vendors selling Outlook OAuth-token accounts Other Always present Recall-card listings say about 0.02–0.03 yuan per account (unverified) Buy finished accounts directly, no need to run your own proxies or beat the human check - Where users come from
- Verified linux.do open-source promotion posts (companion tools posted in rapid succession): the author daimon1 posted Easy Proxies (7/24–7/26; 952 views / 62 likes), GetToken (527 / 23), OutlookManage (424 / 17), and the address generator (2019 / 235); the READMEs link to the linux.do / NodeSeek community pages
- Inferred star-farming / buying stars: 10/21→10/28 lost 522 (41%) in the same window as a site-wide star purge; 11/05 and an unrelated address repo both gained +164/+190 in sync, while the companion repos didn't move; the author 7/25 posted asking whether stars on a resume carry weight, arguing 500+ is worth writing down; there's a github-star-mointor repo. We can't verify the starring accounts one by one.
- Verified forum reputation and demand threads: on a linux.do thread asking for a Hotmail registration bot, someone recommended "find OutlookRegister on GitHub and modify it", while also saying the success rate is low
daimon3332/OutlookRegister research report
Channel: GitHub | Research date: 2026-11-06 | Data: our github database (repo_day), GitHub API, linux.do, NodeSeek, Exa / Jina search, x-radar
Bottom line
A Python script that uses browser automation to bulk-register Outlook/Hotmail accounts and capture Graph OAuth2 refresh_tokens. It's a fork of LainsNL/OutlookRegister, with commits only between 2026-07-23 and 07-27. A large share of its stars were bought, not earned: between 10/21 and 10/28 GitHub's site-wide cleanup of fake star accounts removed 522 of them (41%); between 11/05 and 11/06 it surged in sync with the same author's address repo, while the author's other companion repos didn't move at all. The author has also posted on linux.do asking how much GitHub stars count on a resume (as a new grad). There is real demand (AI account registration bots need large numbers of email addresses), but its star count doesn't show the product is strong, only that someone was pushing stars.
1. What it is
- Uses patchright (an anti-detection Playwright fork) to run multiple browsers that batch-register @outlook.com / @hotmail.com accounts, get past Microsoft's "press and hold" human verification (fully automatic, semi-automatic, or manual mode), grab a refresh_token via Graph OAuth2 after registration, and append results to a txt in the format
email----password----client_id----refresh_token. This is the standard output format of the registration-bot and account-reseller crowd. - The README pushes three companion tools hard: Easy Proxies (a multi-port proxy pool), OutlookManage (an account-management WebUI), and Outlook-Oauth-GetToken (for swapping tokens on existing accounts). Together they form a complete production line: proxy → register → get token → manage accounts.
- Commit history: 8 commits, all between 2026-07-23 and 07-27; apart from the first, all are documentation edits (adding linux.do / NodeSeek community links, resizing icons). No updates since (GitHub API).
- Issues: users report "can't get past the human check," "verification code won't pass," "device or network settings are blocking you to continue" (#1, #15, #16, #17, and others); the maintainer has not replied to any of them (GitHub API, 2026-11-06).
2. Star curve: evidence of inflation
What our github database (repo_day) recorded:
| Date | OutlookRegister | address (same author) | easy-proxies (same author, heavily promoted in README) | OutlookManage (same author) | LainsNL upstream |
|---|---|---|---|---|---|
| 10-13 | 1284 | 2215 | – | 127 | – |
| 10-21 | 1268 | – | – | 136 | – |
| 10-28 | 746(−522) | 1704(−511) | 202 | 137 | 1334 |
| 11-04 | 782 | 1737 | 204 | 142 | 1344 |
| 11-05 | 946(+164) | 1927(+190) | 205 | 142 | 1347 |
| 11-06 | 1022(+76) | 2020(+93) | 206 | 142 | 1350 |
| 11-06 later (GitHub API) | 1112 | 2132 | 209 | 142 | 1353 |
- 10/21→10/28 is GitHub's site-wide purge of bought stars: in the same window, a batch of repos in our database lost 40%–100% of their stars (e.g. edenfunf/reelmimic 3213→242, RubinLabs26/ISpotify 1897→6, feder-cr/git-ai 2579→1011). OutlookRegister lost 41%, address lost 23%. The stars that were removed were the fake accounts' stars.
- The "rebound" for 11/05 is not organic traffic: two unrelated repos (the registration bot and the address generator) both rose 160–190 on the same day; easy-proxies, which the README says is "required", rose only 2 in two days, and OutlookManage rose 0. If people were really drawn in by the registration bot, the companion repos would move with it. The upstream LainsNL gained about 3 stars a day in the same period. Among the repos that got purged, very few kept surging in the following days (ThinkWatch-Lite +97, Motus2 +43; the rest were mostly 0).
- We can't see who is starring: the stargazers endpoint returned 404/403 for our token, so we did not verify the quality of the starring accounts. This point is an inference based on the synchronized rise-and-fall pattern above.
- Motive: on 2026-07-25, two days after this repo was created, the author's linux.do account daimon1 posted "How much do GitHub stars count if you put them on your resume?", calling himself a new grad. In the replies he said "500+ stars should still be fine to list" (https://linux.do/t/topic/2650329). He also has a repo called
github-star-mointor(created 2026-07-24). - Forks have stayed steady at 107–110 and didn't drop during the cleanup, so forks are mostly real. Roughly 110 forks / 1100 stars = 0.1; the upstream is 382 / 1353 = 0.28. The share of people actually using it is far below the upstream's.
3. The real demand and where users come from
- The demand is real: in 2026, the Chinese "registration bot" community is very active. ChatGPT/Codex, Grok, Kiro (AWS), and various API relay sites all need email addresses registered in bulk to harvest free credits, and Outlook/Hotmail is the most common email source. The author himself has forked gpt-free-register and reg-factory and also wrote grok-register. linux.do has many threads like "is there an open-source Outlook registration bot", "registering an Outlook account is so hard", and "Outlook registration bot stuck on human verification" (https://linux.do/t/topic/2233779, /1792503, /2156683); NodeSeek has the same kind of posts (https://www.nodeseek.com/post-650415-1, post-856277-1).
- The organic channel is linux.do's "open-source promotion" section: the author posted several threads in quick succession, 7/24–7/26: Easy Proxies, "multi-port proxy pool, a must-have for registration bots" (952 views / 62 likes); Outlook-GetToken (527 / 23); OutlookManage (424 / 17); and the address generator (2019 / 235, the most popular). OutlookRegister doesn't have its own post in his list (linux.do only shows his most recent 30 posts, not fully verified). The README adds linux.do and NodeSeek "community recommended" links, which is what linux.do's open-source promotion section requires.
- The scale of these posts doesn't explain 1000+ stars: posts with a few hundred views can't account for a star count the upstream didn't reach in 3 months.
- Media and X: searching x-radar for "OutlookRegister" returned 0 posts in the last 120 days; Exa only found auto-indexed sites like DeepWiki and buildradar. No media outlet or major account has written about it.
- Someone replied on a linux.do thread recommending "find OutlookRegister on GitHub and modify it", so there is word-of-mouth, but the same thread also says "the success rate is a bit low".
4. Competitors
| Name | Where | Timeline | Scale | Difference |
|---|---|---|---|---|
| LainsNL/OutlookRegister | Open source | 2025-06 | 1353 stars / 382 forks, 10/31 still being updated | Upstream of this repo; this repo adds secondary-email binding, multi-stage login challenges, cookie-first OAuth, batching, and interruption summaries |
| lxf746/outlook-auto-register | Open source | 2026-08 | 1157 stars, did not drop during the cleanup and instead kept rising steadily (937→1157) | Pure protocol (Fluent Web API), no browser, faster and less resource-intensive; its star growth looks more organic |
| V-IOLE-T/hotmail-register-extension | Chrome extension / open source | 2026-04 | 325 star | Browser extension form, targeting OpenAI registration |
| Older English projects like MatrixTM/OutlookGen | Open source | 2022– | 441 stars, etc. | Old English-language staples, mostly outdated |
| twfb/OutLookEmailRegister, rain-fly, Iron-Yzg, xingluoyuankong, and others | Open source | 2020–2026 | Dozens to 144 stars | Small Chinese projects, all kinds of registration + RT pipelines |
| Wholesale vendors selling Outlook OAuth-token accounts | Other (card-based sellers) | Always present | Recall-card listings say about 0.02–0.03 yuan per account (unverified) | Buy finished accounts directly; on linux.do someone said "a few cents each, why bother with a registration bot" |
5. Differentiation
- The functional difference: a browser-based approach, refresh_token output right after registration, optional secondary-email binding, trilingual READMEs (Chinese Simplified, English, Traditional), and a ready-made proxy pool and account-management tools, so the whole chain is off the shelf. It's a difference in "all-in-one bundle plus good docs", not an algorithmic breakthrough. Getting stuck on human verification is a common complaint in the Issues, and the pure-protocol lxf746 is technically ahead.
- This differentiation is not the main reason it has so many stars. The organic part comes from linux.do promotion plus the registration-bot demand; the bulk of the star count comes from artificial star-pumping (see section 2).
6. How visible it is
Word on the street: it's known in the registration-tool circles on linux.do and NodeSeek, with posts getting a few hundred views. No media, no X coverage. Zero revenue (open-source scripts, not sold).
7. Copycats
I didn't find anyone directly copying it. New forks still show up daily (lilisore, NevinXuHui, and 11/06 other new forks, each with 0 stars). Similar registration tools keep appearing (jiao1yin2he3/outlook-register 2026-09, taskbu 2026-08, etc.), but I can't tell whether they're modeled on it.
8. What to borrow
- Cold start in the Chinese developer community (linux.do's "Open Source Promotion" section): one release of a set of tools that cross-link each other (proxy pool → registration tool → token retrieval → account management), with READMEs linking to each other.
- Cautionary tale: bought stars drop massively during GitHub cleanups (this one lost 41%), and the synchronized rise-and-fall pattern is easy to spot. When judging a project by stars, first compare whether the author's other repos, upstream, and companion repos moved together, then check the fork/star ratio.
- Demand signal: "AI account registration tool → emails, proxies, address data" is a pipeline people keep asking for. A cleaner entry point is tooling like proxy pools or account management (the same author's easy-proxies post got 62 likes).
9. Uncertain
- Star account quality can't be verified one by one (the stargazers endpoint was refused). The star-farming conclusion rests on inference from curve shape, though the evidence is strong.
- How many of the 10/21 stars before 1268 were organic: after cleanup, 746 remain, which sets the upper bound, though some fake accounts may still be left.
- Actual success rate of the tool today: the Issues are full of complaints about failing human verification, and how well it works wasn't tested.
- The wholesale price of token-verified accounts comes from a recall card and hasn't been verified.
Sources
- GitHub API: repos/daimon3332/OutlookRegister (README, commits, issues, forks), users/daimon3332/repos, LainsNL/OutlookRegister, lxf746/outlook-auto-register, 2026-11-06
- Our github repo repo_day, 2026-10-09 to 11-06
- linux.do: /t/topic/2650329 (asking whether stars belong on a resume), /2647587 (Easy Proxies), /2654774 (GetToken), /2655051 (OutlookManage), /topics/created-by/daimon1; demand threads /2233779, /1792503, /2156683, /1920065, /2991003
- NodeSeek: post-650415, post-856277, post-905528
- x-radar search for "OutlookRegister": 0 results