# A journalist handed their grandfather's birth registry to Claude; five generations, not a single error. A Czech genealogy Skill gained 500 stars in three weeks · GitHub open source project

> Source: https://stringers.io/en/github/gh_sarsonj__rodokmen-skill.html (Stringers / 什么值得做, generated 2026-11-08)
> Stringers answers one question every day: what small product can one or two people build in a few months and actually make money from. Numbers come from 15 databases we collect daily, each with source and date.
> For agents: these verdicts and the underlying databases (verified revenue, App Store / Google Play charts, ad libraries, GitHub, Product Hunt…) are queryable live over MCP — tools today / hot / products / directions / direction / data. Setup: https://github.com/komako-workshop/stringers (the user logs in at stringers.io → "配置 agent" to get a personal URL). Site index: https://stringers.io/llms.txt

> The author wrote a download tool for every Czech archive, so even a layman can work through the whole thing; each record comes with its original scan, and Živě.cz only dared write "not a single error" because genealogy veterans checked it. Add a big Facebook group and Czech media coverage in a row, and it became a national story.

[GitHub open-source project](https://stringers.io/en/github/index.md) · written 11-6

sarsonj/rodokmen-skill · sarsonj

# a reporter handed Claude his grandfather's birth record, and five generations came out right: a Czech genealogy Skill, 3 weeks, 500 stars

**What it is** A Czech-language Claude Code Skill. You tell it about your family, and Claude goes into Czech archives on its own, searches the old parish registers, reads old handwriting and Latin, attaches scans of the originals, and builds a family website you can send to relatives.

**Why it took off** The author wrote download tools for every Czech archive, so a non-expert can finish the whole job. Every claim comes with a scan of the original, which is why genealogy veterans at Živě.cz could write "not one error." Add a big Facebook group and Czech media coverage, and it became a national story.

**A** **How you could copy it** Pick a country with good digitized parish archives and no such Skill yet (Poland, Hungary, Italy, etc.), and follow the same structure: interview, one downloader per archive, scans of originals, family website. Build in rate limits from day one, launch in your country's biggest AI Facebook group, and get a local tech journalist who does genealogy to test it.

A grade: Czech version 500 stars, the post-Soviet version 2,000 stars in two weeks. Country-by-country copies work, and most countries are still empty. The bets are whether your country has a media outlet or big group that will spread it, and whether archives will block automated access soon. Buzz in the niche

Data (updated daily) · Repo created 2026-10-16 · 522 stars · 7 days +191

GitHub · stars

Data 2026-10-22 → 2026-11-08 (18 days, 17 observations). Points before 10-28 come from deeply / archive backfill, not every day has one

sarsonj/rodokmen-skill is a Czech-language genealogy Skill for Claude Code. You describe your family, and it searches the parish registers digitized by Czech national archives. It reads German handwriting and Latin from two or three centuries ago, and builds a family website you can share with relatives. Created in Oct 16; by 11-06 it had 502 stars and 79 forks. Most of the stars came from Czech Facebook groups and tech media, and most of the people reading that coverage don't write code. It works not because the Skill is clever, but because the author went through one country's archives one by one. Other countries can copy the approach, and most still have open slots.

### "I shouted at the screen, she's right in front of you"

On Oct 27, 2026 at 8:37 pm, the long-running Czech tech site Živě.cz published an article by Marek Lutonský titled "I gave an AI my grandfather's birth record. Claude built the family tree on its own, five generations, not one error."

Lutonský is no outsider. He has been building his own family tree for over ten years. In the article he says he heard a lot of praise for the tool but didn't believe it, because he knows how much time tracing a tree through archives takes, and how unreliable AI was on this just a little while ago. So he skipped the opening Q&A and gave Claude only a photo of his grandfather's birth record, asking it to find his direct ancestors.

"This was the last specific information I gave the AI." After that he helped with only one thing. The Moravian Archive in Brno (MZA) has strong anti-scraping measures, so once Claude reached the pages, he had to save images under the filenames it specified. At one point he watched Claude search a register index for a woman named Anna Marie Kulhánková and fail to find her. "I nearly shouted at it, she's right in front of you," but he held back and gave no hint. In the end Claude didn't find her either. His family's home village, Vizovice, has many people surnamed Lutonský, and Claude still matched five generations, one by one.

That same night he posted on X: "Another AI WTF moment... I didn't believe it, I was stunned." The repo gained 73 stars the next day and 76 on the third day, the biggest single-day count so far (per the GitHub daily star history).

### What it is

A Skill is a set of operating instructions for Claude Code, plus a small set of Python tools. The README opens clearly: it is for "people who've never done genealogy and can't code," with "Claude handles the technical work, you provide the family details and make the decisions."

- **Start with a chat** : it asks about parents, grandparents, and great-grandparents you remember, going back about a century, because Czech birth records only become publicly available after 100 years.

- **Write tools for each archive** : ARON in Zámrsk, Porta fontium in Plzeň, eBadatelna in Prague, Prague City Archives' police registration records for 1850–1914, Austria's Matricula, plus ANNO and Kramerius for old newspapers. Each has a ready download and search script. The few that don't allow downloads (MZA Brno, Třeboň, Opava) have Claude open a browser with Playwright and page through like a person would.

- **Check everything against the source** : each fact comes with a scan of the original, a transcription and translation, and a confidence level. Searches that found nothing are logged in a search journal too. Opus serves as the main model for review, and Sonnet sub-agents do a first read.

- **Build a final website** : the homepage has family stories, an ancestor fan chart, and a map, with a card for each person, deployed to Cloudflare. A gate asks questions that only family would know to keep outsiders out. The demo site is a fictional Novák family with 260 ancestors across 11 generations.

- **The barrier is money, not skill** : the README says the Pro plan "will most likely not be enough, only good for trying it out," and that a full family tree needs at least Max.

The author, Jindřich Šaršon, founded TappyTaps, a Czech baby monitor app company. This GitHub account has only this one repo and 20 followers. The README's first paragraph says "this does not replace a genealogist," and asks professionals to check its work. In just over two weeks there have been a dozen-plus releases, the latest being v0.9.0 (11-06).

### How the stars came

| Timeline | Stars | What happened |
|---|---|---|
| About 10-16 | 0 | The author had been running an experiment on building family trees with Claude Code in his weekly "Friday roundup." When people asked about it, he refined the workflow on his own family tree and his wife's, then packaged it as a Skill. |
| 10-20 | — | Posted to Czech's largest AI Facebook group, "Umělá inteligence - AI, ChatGPT, OpenAI": 462 reactions, 146 comments, 80 shares |
| 10-22 ~ 10-27 | 59 → 91 | The Facebook wave added a few stars a day |
| Night 10-27 | 91 | Živě.cz published Lutonský's hands-on test |
| 10-28 / 10-29 | 164 / 240 | +73, +76 |
| 10-30 ~ 11-02 | 273 → 374 | Tiscali ran "Two evenings, five generations"; the German genealogy site genealogy.net forum relayed it; Živě ran a follow-up on 11-01; tech personality Patrick Zandl called it "the viral topic of the internet this weekend" |
| 11-03 ~ 11-06 | 435 → 502 | +11 to +60 per day, still not dropping |

(GitHub daily star history, 11-06; Facebook post numbers come from the post pages.)

The repo gained 229 stars in the last 7 days, versus 214 in the first 7 days. The growth didn't end after one news cycle.

Stars are only part of it. Most people in the Facebook group and Tiscali's readers don't have GitHub accounts, and people who installed it and used it wouldn't star it. The clearer evidence is the heavy use that angered archives: 10-30 someone opened an issue saying his agent sent MZA Brno about 40 requests and got his IP blocked. The README now says SOA Praha, Třeboň, and MZA stop letting download tools in starting November. On Okoun, a Czech genealogy forum, longtime members blamed the developer for "messing things up" for the whole research community. The author then added rate limits and stop rules. Repos named "rodokmen-[surname]" have also appeared on GitHub, each one a family website built with it.

### Why this one, and not the generic genealogy Skills

Generic genealogy Skills already exist. emaynard's has 126 stars, sliday's has 39, and mattprusak's autoresearch-genealogy borrowed Karpathy's name for 1,191 stars, but they offer only methodology and templates. rodokmen does three things more:

**First, it knows one country's archives inside out.** Others tell you to "search the parish registers." This one does the searching for you. Which archive uses which interface, and which needs a browser, is built into the tools. An outsider would need weeks to figure that out from scratch.

**Second, its sources can be checked, so experts can vouch for it.** The biggest doubt genealogists have about AI is that it "confidently misreads." Every claim comes with a scan of the original and a transcription, so Lutonský could check them one by one and write "not one error." The "zero errors" line in the coverage comes from that design.

**Three. The finished product is a website you can send to family.** Family trees are made to be shown. On LinkedIn, someone posted "I threw it some old records at night and in the morning it had dug up things from the old registers." A Slovak posted "one weekend, 9 generations, 300 years." Every finished result gets shared again.

Distribution also went in the right order: first the country's biggest AI Facebook group, then waiting for local tech journalists who know genealogy to test it. Telling real family stories in the local language works far better than a cold start on GitHub or HN. The repo is basically unknown outside Czechia; searches on HN and Reddit turn up nothing.

### Who else is in the space

| Project | Stars | Compared with alternatives |
|---|---|---|
| DenisSergeevitch/ex-ussr-kinship-research | 2,025 | 10-25 built, same region-by-region archive genealogy approach, covers the post-Soviet region; the author has 1,202 followers; no website generated |
| mattprusak/autoresearch-genealogy | 1,191 | generic prompts, Obsidian templates, archive guides; no per-country download tools |
| emaynard/claude-family-history-research-skill | 126 | research planning only |
| vzoha/dejiny-domu-skill | 63 | derived from rodokmen, changed to search house and village history |
| midlan/skill-czech-genealogy | — | Czech and Slovak; tools carried over directly from rodokmen (MIT) |
| sliday/genealogy-research, mistok-wiki/chytalnia (Ukrainian villages), etc. | dozens | generic methodology, or only method with no tools |
| AI features from MyHeritage, Ancestry, FamilySearch | Tens of millions of users | only searches your own paid databases; weak coverage of Czech digital archives |

(GitHub, 11-06.)

The first two rows matter most: **the same pattern gave Czech 500 stars, and the post-Soviet version 2,000 stars in two weeks.** Country-by-country works, and star counts track the size of the population and language, plus whether the author already has an audience.

### If it were you

**A: copying the pattern to another country has a clear recipe, proven by both the post-Soviet and Czech versions. The bets are two: whether your country has a media outlet or big group that will spread it for you, and whether the target archives block automated access soon.**

**Who it suits** : developers whose native country has well-digitized parish archives but no such Skill yet, ideally people who have done genealogy themselves or with family and can read old registers. Better still if you have a voice in that country's AI community, genealogy circles, or tech media.

**Who it doesn't suit** : anyone aiming for an English general-purpose version (a generic methodology repo with 1,000+ stars already holds that ground, and without country-specific tools there's no differentiation); anyone wanting another Czech or Slovak version (rodokmen and midlan already cover that); anyone unwilling to deal with each archive's interface and anti-scraping, who only wants to write prompts.

**How to do it** : three ways to follow, best to worst:

1. **Switch countries** : parish archives in Poland, Hungary, Italy, Croatia, and Slovenia are heavily digitized (Matricula alone holds many parishes from Austria, Germany, Poland, and Slovenia), and I haven't found matching Skills (not verified one by one). Follow rodokmen's structure: interview → one downloader per archive, browser fallback for those that block it → scans of originals plus transcriptions and confidence levels → family website. It's MIT-licensed, so you can reuse the tool skeleton directly.

2. **Switch data sources** : instead of people, search other public archives, such as house and village histories (vzoha is already doing this), land registries, family news in old newspapers, or emigrant ship lists.

3. **Fix what it got wrong** : it hit archives too hard, and several have already blocked it. A version that builds in rate limits, caching, and per-user daily page caps from the start, and reaches out to the archives proactively, could be its own story.

**How to get users** : follow rodokmen's order. Run it on your own family tree first, then share the finished website as an example. Post in your country's biggest AI community (Facebook groups, Telegram channels, local forums), telling your own family's story rather than the tech. Then find local tech journalists who've done genealogy and ask them to test it on their own old records. Copy the README's first-paragraph stance: it doesn't replace a genealogist, and feedback is welcome.

**When to move** : this wave of country-specific archive genealogy Skills only started in late October. The post-Soviet, Czech, and Czech-derived versions all appeared in the same month. My guess is the popular countries will be taken within a few months, so the best window is before Q1 2027 (inference). Christmas, when families gather and talk about ancestors, is a natural moment to spread it (inference).

**Signs you've misread it** : ① you post in your country's big AI group and it doesn't reach 100 stars in a week, with no media asking about it, which would mean the topic is less compelling outside Czechia and the post-Soviet region; ② the main archives in the country you pick start blocking automated access within a month or two, leaving only the browser route, which makes the experience much worse and stalls growth; ③ the Claude subscription threshold (at least Max) keeps casual readers from trying it even after the news, so stars and discussion come only from developers, and you shouldn't expect it to spread beyond that circle like rodokmen did.

[GitHub repo ↗](https://github.com/sarsonj/rodokmen-skill)

**Research notes: competitor table, acquisition evidence, full research report**

Competitors

| Name | Where | Launched | Scale | How it differs from this one |
|---|---|---|---|---|
| DenisSergeevitch/ex-ussr-kinship-research | Open source | 2026-10-25 | 2025 star | A region-by-region archive genealogy Skill in the same wave, covering the post-Soviet region; the author has 1200 followers; it doesn't generate family websites |
| mattprusak/autoresearch-genealogy | Open source | 2026-03-18 | 1191 star | General prompts, an Obsidian template and an archive guide; no country-specific archive download tool, and no finished website |
| emaynard/claude-family-history-research-skill | Open source | 2025-10-17 | 126 star | Research planning only, does not query registers automatically |
| sliday/genealogy-research | Open source | 2026-03-23 | 39 star | GPS methodology + Obsidian, general purpose, aimed at researchers |
| The AI features of MyHeritage / Ancestry / FamilySearch | Big-company features | Years | Tens of millions of users | Only searches its own paid libraries; Czech digital archive coverage is weak. rodokmen goes straight to the free national archives |
| Transkribus | Web | Years | Big | Handwriting recognition only, no search, no reasoning, no family tree building |
| Czech human genealogy services (vasegeny.cz, rodokmenprofi) | Other | Years | Small operations | Human-done and paid; rodokmen only needs a Claude subscription, so laypeople can do it themselves |

Where users come from

- **Verified** **The author posted in the big Czech AI Facebook group** : 2026-10-20 posted to the "Umělá inteligence - AI, ChatGPT, OpenAI" group, 462 reactions, 146 comments, 80 shares; he had teased it earlier in his Friday roundup

- **Verified** **Tech media hands-on review** : Živě.cz ran 2026-10-27 20:37 Marek Lutonský's hands-on piece "Five generations with zero errors"; the author posted it to X (@marekl) the same evening; from the next day stars rose by +73 and +76

- **Verified** **Follow-up coverage and sequels** : Tiscali 10-30, ejaj.cz reprint, Živě 11-01 sequel, German forum genealogy.net 10-31 relay

- **Verified** **Big X tech account** : Patrick Zandl @tangero 11-01 called it the "weekend internet viral topic"

- **Verified** **Users showing off results (LinkedIn, Instagram, forums)** : On LinkedIn, Jan Janouškovec and Karin Kramara (9 for 300 years) shared their results; Instagram scrolls; discussion on the Okoun genealogy forum

- **Verified** **Skill aggregator sites** : skillleaderboard.com and claudepluginhub.com index it automatically

## sarsonj/rodokmen-skill: a Czech-language Claude Code skill that builds family trees automatically

### One line

A Claude Code skill/plugin written by Czech developer Jindřich Šaršon (founder of TappyTaps, the company behind a baby monitor app). Users talk about their family, and Claude goes off to search the digitized parish registers (matriky) of Czech state archives: it reads Kurrent handwriting and Latin, transcribes and translates, attaches scans of the originals, and generates a family website (ancestor fan chart, maps, family stories) deployed to Cloudflare, gated by a "family question" as the access password. The README is entirely in Czech and covers only Czech and Austrian archives.

### Data (GitHub repo_day + GitHub API, 2026-11-06)

- Created 2026-10-16, 21 days ago, 502 stars (the API shows 508 today), 79 forks, 25 open issues; between 10-16 and 11-06 about 10 releases shipped (v0.5.0 to v0.9.0).

- Curve: 10-22 59, 10-27 91,10-28 164 (+73), 10-29 240 (+76, the single-day peak), and then +25 to +60,11-06 502 a day. gain_7d 229; the first 7 days 214, and it hasn't come down yet.

- The author's account has only this one repo and 20 followers; he has no influence on GitHub of his own.

### Why it took off: timeline (every point is sourced)

1. **About 10-16: the author wrote about experimenting with building family trees using Claude Code in his own "Friday roundup"** . When people asked, he tuned the workflow on his own family, packaged it as a skill, and then tried it on his wife's family too (quoted from the author's FB post).

2. **2026-10-20: the author posted to Czech's biggest AI Facebook group, "Umělá inteligence - AI, ChatGPT, OpenAI - sdílení zkušeností"** , getting 462 reactions, 146 comments and 80 shares (https://www.facebook.com/groups/umelainteligence/posts/2604029143394595/ ). This was the first wave; by 10-27 it had reached about 90 stars. Most of these people aren't developers, and many would never go to GitHub to star anything, so stars badly understate the real spread.

3. **2026-10-27 20:37: Živě.cz (a long-running Czech tech outlet) ran Marek Lutonský's hands-on piece** , titled "I gave AI my grandfather's birth record, and Claude built a five-generation family tree on its own, with no errors" (https://www.zive.cz/clanky/claude-code-skill-rodokmen/sc-3-a-244120/default.aspx ). The author, who has been doing genealogy for over ten years, posted it to X (@marekl) at 2026-10-27 20:49 UTC the same night ("another AI WTF moment... I don't believe it, I'm stunned"). The following day and the day after were the biggest peaks, +73 and +76, and the timing lines up.

4. **Spread to mainstream media and beyond the niche** : Tiscali 10-30 wrote up "Five Generations in Two Evenings" (https://nedd.tiscali.cz/pet-generaci-za-dva-vecery-... ), reporting it at "over two hundred stars"; ejaj.cz reposted it; the German genealogy community genealogy.net relayed it in a forum 10-31; Živě 11-01 ran a follow-up, "I Built an AI Genealogist for Myself," opening with "the last post drew a lot of interest on the Czech internet"; Patrick Zandl (@tangero, a big name in Czech tech circles) 11-01 called it "the weekend internet's viral topic." There was also Instagram scrolling, personal LinkedIn posts showing off results (one Slovak user: "five generations, 300 years, in one weekend" — 9), and discussion on the Okoun genealogy forum.

5. **Retention and word of mouth ran on personal stories of actually finding ancestors** : On LinkedIn, Jan Janouškovec wrote, "I toss it the task in the evening, and in the morning I see what it dug up in the old registers"; Zandl found a great-uncle who had been director of the Chirana factory. The finished output is a website you can send to family, which naturally gets shared.

**Takeaway** : this is a **localized, non-developer-friendly Claude Code skill that first broke out in Czech Facebook groups, then was amplified by Czech tech media's hands-on reviews** . The GitHub stars were a byproduct of this nationwide wave.

### Differentiation

- **Mastering one country's archive infrastructure** : Python download and search tools were written for ARON (Zámrsk), Porta fontium (Plzeň), eBadatelna (SOA Praha), AHMP police registers (1850–1914), Matricula, and ANNO / Kramerius newspapers. Archives where the download tool can't get in (MZA Brno, Třeboň, Opava) fall back to a Playwright browser. Generic genealogy skills offer only methodology, with none of these ready-made interfaces.

- **The full workflow handed to ordinary people** : from interview to website deployment, end to end. The README states plainly that "you can use it even if you can't code and don't know Latin or Kurrent," and it prescribes a model combination (Opus as the main model for review, Sonnet sub-agents for first reads).

- **Verifiable** : every fact comes with a scan of the original, a transcription and a translation, plus a confidence rating, and it keeps a search log (including searches that found nothing). This directly answers the genealogy community's objection that "AI confidently misreads things," which is why the Živě reviewer, a ten-year veteran, was willing to write "zero errors."

- **The author is humble and iterates fast** : the README opens with "does not replace a genealogist," invites professionals to point out errors, and shipped about 10 versions in two weeks.

Its differentiation is why it took off: **"a Czech-archive-specific tool that generates a website you can send to family"** lets laypeople complete, end to end, something that used to take years of study. Contemporary generic genealogy skills (emaynard, sliday, etc.) have only dozens to a hundred-odd stars.

### Competitors

| Name | Where | Launched | Scale | How it differs |
|---|---|---|---|---|
| DenisSergeevitch/ex-ussr-kinship-research | Open source | 2026-10-25 | 2025 stars (author has 1202 followers) | The same wave of "archive genealogy skills by region" covering former Soviet regions; these have authors with their own followers and more stars; they don't generate family websites |
| mattprusak/autoresearch-genealogy | Open source | 2026-03-18 | 1191 star | General prompts, an Obsidian template and an archive guide, borrowing the name of Karpathy's autoresearch; no country-specific download tools |
| emaynard/claude-family-history-research-skill | Open source | 2025-10-17 | 126 star | Research planning only; doesn't download registers or produce websites |
| sliday/genealogy-research | Open source | 2026-03 | 39 star | GPS methodology + Obsidian, general purpose |
| mistok-wiki/chytalnia | Open source | 2026-09 | 27 star | A method for Ukrainian village archives |
| vaelen/gedcom-skills, tamvodopad/… and others | Open source | 2026 | Single-digit stars | Processes GEDCOM files, doesn't touch archives |
| The AI built into MyHeritage, Ancestry and FamilySearch | Big-company features / web | Years | Tens of millions of users | Searches only within their own libraries and charges a subscription; Czech digital archive coverage is weak; they deliberately don't bulk-scrape these paid sites |
| Transkribus | Web | Years | Big | Handwriting recognition only, no search, no reasoning, no family tree building |
| Czech professional genealogy services (vasegeny.cz, rodokmenprofi, etc.) | Other | Years | Small operations | Human-done, costing thousands to tens of thousands of crowns; rodokmenprofi has posted a video responding that "our way of using AI is different, legal, and not in violation" |

### Copycats / derivatives

- vzoha/dejiny-domu-skill (10-30 created, 63 stars): forked from rodokmen v0.5.0 and retooled to research "the history of houses and villages"; covered in the Živě sequel.

- midlan/skill-czech-genealogy (10-25): Czech + Slovak with Gramps, with tools moved over directly from rodokmen (MIT).

- micl79/rodokmen-skill (revised version), zbysekvoda/rodokmen-plus, thejanhromek/pamet-rodu (a single-HTML family chronicle), pe-pan/matriky-index, musilrad13-rgb/MatrikyAI; in Russian, GuruDevvv/family-tree-starter (AI assistant + family tree website + maps, a very similar shape); Marek Lutonský's own "Genealog" variant.

- Plus a large number of individual "rodokmen-<surname>" repos, each a finished product built by a user with it.

### Side effects (worth noting)

So many people used it that Czech archives began blocking automated access. Issue #45 (2026-10-30): one user's agent sent about 40 requests to MZA Brno in quick succession, and the IP got banned. The README now says "SOA Praha, Třeboň and MZA have refused download tools since 2026 November; browser only." On the Okoun genealogy forum, longtime users blamed the developer for "messing things up" and dragging down the whole research community. In v0.6–0.7 the author added rate limits and stop rules. This shows a fair number of people really are using it, but the ceiling on this approach is archives' anti-scraping defenses.

### Visibility

It's already well known inside the Czech scene: two pieces in national tech media, one in a portal, shares from a major tech account, several hundred reactions on the big FB group. Outside the Czech Republic almost nobody knows about it; only genealogy.net relayed it, and aggregators like skillleaderboard and claudepluginhub index it automatically. No revenue, and it's free (MIT; the only requirement is that users have a paid Claude subscription).

### What to take from it

- **"Claude Code skill × a country's public archive or data portal × a non-developer's emotional need"** : find a nationally digitized public data source with poor tooling (parish registers, land registry, old newspapers), and package a downloader, reading guidelines and a finished website as a skill. The Soviet-region version got 2000 stars, showing the pattern can be copied country by country (parish archives for Poland, Slovakia, Hungary and Italy are still empty).

- **Distribution** : the author first posted to his country's biggest AI Facebook group, then waited for local tech journalists to come test it. Local language plus real people's stories works far better than a cold launch on GitHub or HN.

- **Verifiable design** : scans, transcriptions and confidence labels are the key to winning over professional skeptics, and the reason media dared to write "zero errors."

- **Mind the ceiling** : target data sources fight back against scraping, so rate limiting and compliance have to be built in from the start.

### Uncertain

- How many people actually installed it: Facebook and plugin install counts are unavailable, and stars are clearly understated.

- The Živě first piece's readership wasn't obtained (the page was blocked by Cloudflare; the body text came from Exa).

- How much the archive bans reduce the experience for new users: not verified.

- Whether there's a HN or Reddit thread: none found in search.

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