# One landing-page compilation video took a Markdown file from 2.8k to 10k stars in 4 days: designer Elaya's AI design skill · GitHub open source

> Source: https://stringers.io/en/github/gh_elayadesign__ai-design-skills.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 is a designer taking client work who used it to build a full screen of good-looking real work, cut it into a compilation video that blew up on X, and that same week coincided with a collective surge of "designer-written AI skills", reaching #1 on GitHub's site-wide daily chart.

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

elayadesign/ai-design-skills · elayadesign

# One landing-page compilation video took a Markdown file from 2.8k to 10k stars in 4 days: designer Elaya's AI design skill

**What it is** An open-source Markdown file. Once installed in Claude Code, Cursor, or Codex, it has AI build landing pages following a fixed process and visual rules.

**Why it took off** The author is a freelance designer who used it to build a full screen of polished real work, cut it into a compilation video, and posted it to X, where it blew up. It also landed in the same week as a wave of "designer-written AI skills," reaching #1 on GitHub's all-site daily trending.

**C** **How you could ride it** Stop building generic design skills. If you're a designer with real work, pick a narrow niche (App screenshots, PPT, Chinese landing pages), write the skill, post finished outputs on X for two months, then release a compilation video, and aim for Trending once the repo hits 60 days old.

Grade C: The playbook is clear and learnable, but generic design skills already have four 100k-star projects, and a copycat wave hit the same week, so the wave has peaked. It took off because of the author's own portfolio and taste, which followers can't copy. Only designers with real work doing a narrow niche should try it. Public / widely discussed

Data (updated daily) · Repo created 2026-07-29 · 14,830 stars · 7 days +1,730

GitHub · stars

Data 2026-10-09 → 2026-11-08 (31 days, 30 observations). Points before 10-28 come from deeply / archive backfill, not every day has one

elayadesign/ai-design-skills is an open-source file written by Elaya, a freelance product designer based in Dubai, for Claude Code, Cursor and Codex. It teaches the AI how to build landing pages. In Oct 24, he posted a compilation video titled "Every landing page I've built with it these last two months": 877k views. The repo went from 2,844 stars to 10,187 stars in 4 days and took the #1 spot on GitHub's daily trending chart. The rules themselves aren't especially original; several projects in the same category have 100K+ stars. What made it take off was a designer with real client work showing a whole screen of actual output. The playbook is worth copying, but this lane is already crowded.

### A tweet with 877k views

2026 Oct 24 17:39 (UTC), Elaya (@elayadesigns on X, about 9k followers, over 8 years in product design) posted:

> "I designed all of these landing pages in the last two months, using @higgsfield, Claude, and Cursor with my custom AI designer skill. The skill is open source."

Attached was a compilation video of one animated SaaS landing page after another, with motion and video assets generated in Higgsfield. By Nov 6 the tweet had 877k views. A few minutes later he replied to his own thread with a link to his personal site, "Work With Elaya — Freelance Product Designer for SaaS & Startups." Open source was the business card; freelance work was the goal.

The star changes match the tweet's spread (GitHub daily stars):

| Date | Stars | What happened that day |
|---|---|---|
| 10-24 | 2,833 | Compilation video posted |
| 10-25 | 2,844 | Japanese blogger @tetumemo reposted it with a Japanese breakdown |
| 10-26 | 4,759(+1,915) | Design-world account Tommy Geoco @designertom posted "Designer-made agent skills are everywhere this week," grouping it with taste-skill and hallmark |
| 10-27 | 7,473(+2,714) | Single-day peak |
| 10-28 | 10,187(+2,714) | #1 on GitHub's all-site daily trending |
| 10-29 ~ 10-31 | 11,264 → 12,816 | #2 and #7 on daily trending, with a long tail of reposts on Medium, YouTube, and IG reels |
| 11-06 | 14,292 | Daily forks fell back to 159,1,070 |

The fork-to-star ratio is about 7.5%, which is normal; no sign of star-botting.

### What it actually is

The repo now has two skills: the main `landing-page-design` and a later `figma-designer` (building designs in Figma using variables, text styles, and auto layout). The main one is a 484-line SKILL.md, installed into `.claude/skills/` with one command, `npx skills add elayadesign/ai-design-skills`, or pasted directly into Cursor rules.

The file has two halves. The first governs "what the page says": before starting, settle the key questions once. What's the single conversion action, who's the target user, what are three reasons they wouldn't convert, and where does traffic come from? Then set the hero structure, write conversion copy, and handle SEO. The second half governs "what the page looks like," with very strict rules:

- Fonts are limited to Geist, Manrope, Geist Mono, and Poppins; **never Inter, Roboto, or Arial** . No italics, no 900 weight.

- No hyphens in body text, and a headline's last line may not be a single word.

- Every page needs a large tagline section whose words light up one by one from 30% opacity as you scroll.

All values for fonts, colors, spacing, and motion are gathered in one section at the bottom of the file. The README says "Fork this repo, edit that section," so changing one section gives you your own style. MIT, "No attribution needed."

### Why this one, and why this week

**First, it demos the finished product, not the rules.** Viewers didn't read those 484 lines. They saw screen after screen of beautiful pages they could use directly, then were told "you can get this for free too." Starring it feels like bookmarking "the thing that makes pages like this."

**Second, it didn't appear out of nowhere.** The author tweeted the day the repo was created in Jul 29. From August on, nearly every work-showcase tweet said "Higgsfield + Cursor + my custom landing-page-design-skill" and linked the repo. One tweet got 8-20 81k views, another 8-29 74k. Over two and a half months he built up 2.8k stars. What he built was a group of followers who trusted his taste, and the compilation videos blew up on top of that foundation.

**Third, it rode a wave.** "AI-generated pages all look the same" (Inter font, purple-blue gradients, cards inside cards) was a complaint everyone had this second half of the year, and SKILL.md had become the common extension format for Claude Code, Cursor, and Codex. The last week of October was a collective burst of designer-written skills: in that same week hallmark gained 9.7k stars and impeccable gained 11.8k. The wave didn't lift it up; it was simply the one that got noticed in the wave.

**Fourth, it caught GitHub Trending.** The repo was already 60 days old, old enough to qualify for the all-site chart. Getting #1 on 10-28 brought a second wave of 10-28 to 10-31 stars a day, 700 to 1000 a day after that. This was timing, not planning.

Naming @higgsfield in the copy also helped: Higgsfield itself runs heavy ads across the web, and "I made this with Higgsfield…" posts get reshared within its user community.

### Who else is doing the same thing

| Project | Stars (11-06) | Compared with alternatives |
|---|---|---|
| anthropics/skills (frontend-design, etc.) | 187.7k | Official generic frontend design skill, not focused on conversion copy |
| nextlevelbuilder/ui-ux-pro-max-skill | 145.6k | Design knowledge base with search that gives suggestions, not specialized in landing pages |
| Leonxlnx/taste-skill | 110.5k | Gives the agent "taste": art direction, reference images, browser checks |
| pbakaus/impeccable | 103.8k | 62 deterministic detectors plus CI, 2,622 commits, the heaviest engineering of the group |
| emilkowalski/skills | 53.8k | Motion and interaction polish, applied to existing code |
| Nutlope/hallmark | 48.6k | Renders at 8 widths, then re-reviews in a clean context |
| alchaincyf/huashu-design | 28.8k | Chinese community, outputs HTML, PPT, MP4, PDF |
| Motionsites-AI/claude-skill | 1.7k | Created 10-26, "cinematic landing pages," with 50+ built-in video assets |

In this group, elaya has the thinnest engineering: 16 commits, no detectors, no render verification. What it has uniquely is just two things: it does one job, landing pages (it even handles requirements questions and copywriting), and the author's own portfolio serves as the demo.

There's also a side story. The day after the compilation video went out, taste-skill's author Leon Lin replied directly under the thread: "hey dude, would appreciate some credits for tasteskill," with a few comparison screenshots. The next day elaya added attribution to the README, saying it borrowed content and motion ideas from taste-skill. So part of its rules came from someone else too.

Copycats are already here too: Motionsites-AI/claude-skill was created on 10-26, the same day the trend took off, with the title "Free Claude skill: make Claude build award-level, cinematic landing pages". It hit 1.7k stars in 11 days, and behind it is a motion-asset site using it to drive traffic.

### What the stars brought

Open source itself doesn't charge, and the author's returns come from two places. First, client work: the viral post linked directly to a personal site. Second, SlopScan (slopscan.co), which uses the same rules to audit live pages and generate fix prompts, put in a banner at the top of the README starting Nov 1. On TrustMRR it has earned $14 over the last 30 days. It's just getting started. For people building open source, the lesson isn't the $14; it's the path: a repo with 14k stars has a free ad slot in its homepage banner.

### Its weak spots

- **Thin.** It's one Markdown file. Anyone can copy the rules, and it has borrowed from others.

- **Heat is fading.** Daily adds dropped from 2,714 to 159, back to flat two weeks after the spike. The author 10-25 added design-to-motion, and 10-27 said it would be "moved to their own repo"; no new public repo has shown up since.

- **Most stars are probably bookmarks.** How many people actually installed it and use it isn't visible from outside; there are only 2 open issues.

### If it were you

**Bottom line:** Learn the playbook, but don't build another "AI design skill." Generic design skills already have four 100k-star projects, and more appeared the same week. Another copy would be hard to notice. It took off because of the author's real screen of work, which followers can't reproduce.

**Who it suits:** Designers with a body of strong work they can show, who already have some followers on X or elsewhere. Or developers who are genuinely better than others in a vertical niche (App Store screenshots, email templates, PPT, e-commerce detail pages, Chinese landing pages).

**Who it doesn't suit:** People with no work to demo; people hoping to win by writing more complete rules, since impeccable and hallmark already have the engineering nailed; people hoping to make money from it, since an open-source skill doesn't earn money on its own.

**How to do it:**

1. Pick a narrow niche, not "design." For example, only App Store screenshots, only pitch decks, only Xiaohongshu covers, only Chinese SaaS landing pages. The narrower the niche, the easier it is to become the first answer for that niche.

2. Gather every tunable value (fonts, colors, spacing, motion) into one section of the file, so others can fork it and turn it into their own style. More forks also bring more exposure.

3. Whatever rules you borrowed, write Credits on day one, so nobody has to ask for them in the replies to your hit post.

**How to get users:** Follow elaya's rhythm: post the day you create the repo, then post again with every piece of work, always linking the repo, and build up two months of posts. Once you have a dozen good-looking finished pieces, cut a reel called "Everything I made with X these last two months" and tag the hot tools you used (Claude, Cursor, some generation tool) so their user circles repost it for you. Post this big push on day 60 of the repo, which is just in time to catch GitHub Trending. Then DM the video to repost accounts in Japanese, Chinese, and design circles.

**When was the window:** The wave of "designer-written skills" peaked in late October. The general-purpose category's window is basically shut. Vertical niches still have room, but new agent tools and official skills update fast, so I'd estimate you have a few months, not a year (inference).

**What would mean you read it wrong:** You run the two-month work-post push and each video still gets under a few thousand views, which means the work itself isn't striking enough and this playbook doesn't hold. Or the narrow niche you picked gets an official skill or a submodule in a 100k-star project within a month or two. In that case, stop doubling down on that path.

[GitHub repo ↗](https://github.com/elayadesign/ai-design-skills)

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

Competitors

| Name | Where | Launched | Scale | How it differs from this one |
|---|---|---|---|---|
| nextlevelbuilder/ui-ux-pro-max-skill | Open source | 2025-11 | 145.6k stars | A design knowledge base that combines data and retrieval to give advice; it doesn't focus on landing-page conversion |
| Leonxlnx/taste-skill | Open source | 2026-02 | 110.5k stars | A general "taste" skill with reference images and browser checks; elaya borrowed its content and had to add credit only after being asked |
| pbakaus/impeccable | Open source | 2025-11 | 103.8k stars, 2,622 commits | 62 deterministic checkers plus CI, far more engineered than elaya's 16 commits |
| emilkowalski/skills | Open source | 2026-03 | 53.8k stars | Rules for polishing motion and interaction, applied to existing code rather than generating pages from scratch |
| Nutlope/hallmark | Open source | 2026-04 | 48.6k stars | An anti-slop skill from Together AI, with multi-width rendering and a second-pass review |
| alchaincyf/huashu-design | Open source | 2026-04 | 28.8k stars | Aimed at the Chinese-speaking market, with multi-format output (HTML/PPT/MP4/PDF) |
| anthropics/skills(frontend-design) | Big-company features | 2025-09 | 187.7k stars | The official general-purpose frontend design skill, with no requirements interview and no conversion copy |
| Motionsites-AI/claude-skill | Open source | 2026-10-26 | 1.7k stars | Cinematic landing pages with 50+ video assets, used to drive traffic to MotionSites |

Where users come from

- **Verified** **the author's X portfolio videos went viral** : 10-24 pinned the post "I designed all of these landing pages…The skill is open source", 877k views; stars rose daily from the next day, 1.9k–2.7k

- **Verified** **the author has posted their work on X for a long time** : 7-29 posted the launch; 8-20 one post got 81k views, 8-29 another got 74.4k views, both linking the repo; by early October the total was about 2.3k stars

- **Verified** **riding on hot tool names** : the copy tagged @higgsfield, Claude, and Cursor Composer 2.5

- **Verified** **KOLs and multilingual reposting** : @tetumemo's Japanese breakdown (10-25), @designertom's roundup of design skills (10-26), @neil_xbt (10-27), plus Medium, explainx, YouTube, and IG reels

- **Verified** **GitHub trending** : all-GitHub daily chart 10-28 #1 (+1,395), 10-29 #2, 10-30 #7

- **Verified** **skills directories and SEO** : auto-indexed by skills.sh, ui-skills.com, typingmind, agentskillshub, skillleaderboard, and others

## elayadesign/ai-design-skills: a designer's portfolio video pushed one SKILL.md to #1 on GitHub's all-site chart

Research date 2026-11-06. Channel: GitHub.

### One line

Elaya (X @elayadesigns, Dubai, about 9k followers), a freelance product designer with 8+ years of experience, wrote `landing-page-design`, a Markdown skill for Claude Code / Cursor / Codex. It asks what the client needs first, then sets the page structure, writes the conversion copy, and finally applies a full visual system. Later they added `figma-designer`. The repo went up in July 29, and the author kept posting "landing pages made with this skill" on X, slowly building it to 2.8k stars. In October 24 a portfolio reel got 877k views; within 4 days stars went from 2.8k to 10k+, taking #1 on GitHub's all-site daily chart.

### Data (our github repo_day database; source noted separately)

- 10-09 2,355 → 10-25 2,844 (daily gain 8–85)

- 10-26 4,759 (+1,915) → 10-27 7,473 (+2,714) → 10-28 10,187 (+2,714) → 10-29 11,264 → 10-31 12,816 → 11-06 14,292 (daily gains fell back to 159)

- 1,070 forks; forks/stars about 7.5% (the median for new repos with ≥5000 stars is about 11%); the suspect flag in our database is empty. 2 open issues, 16 commits, no releases.

- GitHub trending all-site daily chart (our database, Beijing time): 10-28 #1 (that period +1,395), 10-29 #2,10-30, #7. star-history.com records 10-27 #1, 10-28 #2, 10-29 #7, for 3 days on the chart in total.

- The author's other repo, redesign-skill (created 7-18, audits and remakes existing websites), has only 520 stars.

### Why it took off

#### Phase one (7/29 – 10/23): the author posts their own work and slowly builds stars

- The launch post went out 7-29 07:04 UTC: "My Landing page design skill, open source intake, structure, copy, and a full visual system in one file" (https://x.com/elayadesigns/status/2082361627282940409), almost at the same time as the repo was created.

- From August on, they kept posting motion landing pages built with "Higgsfield + Cursor Composer 2.5 + my custom landing-page-design-skill", every post linking the repo. 8-20 one post got 81k views (https://x.com/elayadesigns/status/2090356254095130825), 8-29 "Rebuilding my most viral landing page from last month" got 74.4k views (https://x.com/elayadesigns/status/2093739197492605107).

- Result: by early October it already had 2.3k stars (aiagentskills.net recorded 2,310; our database has 10-09 at 2,355), then gained dozens a day.

#### Phase two (from 10/24): one portfolio video sets it off

- **10-24 17:39 UTC** : the author posted a pinned tweet: "I designed all of these landing pages in the last two months, using @higgsfield, Claude, and Cursor with my custom AI designer skill. The skill is open source," with a reel of several landing pages, **877k views** (https://x.com/elayadesigns/status/2114049200749175174). getxanalytics filed it under "AI Coding launch videos."

- In the same thread the author replied with elaya.design, "Work With Elaya — Freelance Product Designer for SaaS & Startups," 43.5k views (per Google's snippet). The open source is there to bring in clients.

- Secondary spread:
    - 10-25 Japanese creator テツメモ @tetumemo broke it down in Japanese and reposted (https://x.com/tetumemo/status/2114314782035837369)
    - 10-26 Design-world account Tommy Geoco @designertom posted "Designer-made agent skills are everywhere this week," introducing it alongside taste-skill and hallmark (https://x.com/designertom/status/2114786186250379759)
    - 10-26 The author posted "reached 7.7k stars," teasing design-to-motion (https://x.com/elayadesigns/status/2114806587466604676)
    - 10-27 @neil_xbt wrote "Their post showing the pages went viral in two days" (https://x.com/neil_xbt/status/2115023701267947654)
    - After that came a long tail of content: Medium (10-28, [medium.com/@mehulpratapsingh/](https://medium.com/@mehulpratapsingh/)...), explainx.ai, thakicloud, codearia, YouTube videos, Instagram reels, LinkedIn reposts

- It reached #1 on GitHub's all-site daily chart, and that brought a second wave (10-28 to 10-31, 700–1000+ per day).

- The timing lines up: on 10-25 one day gained +11,10-26, +1,915,10-27/28 each +2,714, matching the spread of that video.

#### Why people wanted it at that moment

- In the second half of 2026, agent skills (SKILL.md) became the standard way to extend Claude Code / Cursor / Codex. skills.sh installs one with a single `npx skills add` line. "Every page an AI builds looks the same" (Inter font, purple-blue gradients, cards inside cards) is a widely recognized pain point. The setproduct 11-03 long-read calls this kind of skill a tool for fixing "AI slop" (https://www.setproduct.com/blog/design-skills-for-ai-agents-compared).

- The last week of October coincided with a collective surge of "designer-made skills": designertom's words were "everywhere this week"; hallmark gained 9.7k in a week, impeccable +11.8k. It was one of the wave's members, but the wave wasn't started by it.

#### Why people were willing to star it

- The demo shows the result, not the rules: viewers see screen after screen of beautiful landing pages they could use directly (motion, with video assets from Higgsfield), then get told "you can get this same skill for free." A star largely means "bookmarking the thing that makes pages like this."

- The barrier is minimal: a single Markdown file, copied into `.claude/skills/` and it works; MIT license, no attribution required; all design values are gathered in one section at the bottom of the file, so forking and editing that section makes it your own style.

### Competitors

| Name | Form | Launched | Scale (github database 11-06 / setproduct 11-03) | Difference |
|---|---|---|---|---|
| nextlevelbuilder/ui-ux-pro-max-skill | Open source | 2025-11 | 145.6k stars | A design knowledge base that combines data and retrieval to give advice; not focused on landing pages |
| Leonxlnx/taste-skill | Open source | 2026-02 | 110.5k stars | Gives agents "taste": art direction, reference images, browser checks. elaya's skill borrowed its content and motion ideas; 10-25 asked for credit in person from Leon Lin, and only then 10-26 was added |
| pbakaus/impeccable | Open source | 2025-11 | 103.8k stars, 2,622 commits | 62 deterministic checkers plus CI, the most engineered of the group |
| emilkowalski/skills | Open source | 2026-03 | 53.8k stars | Polishing rules for motion and interaction details, applied to existing code |
| Nutlope/hallmark(Together AI) | Open source | 2026-04 | 48.6k stars | Anti-slop: rendered at 8 widths, then reviewed in a fresh context |
| alchaincyf/huashu-design | Open source | 2026-04 | 28.8k stars | Aimed at the Chinese-speaking market; outputs HTML/PPT/MP4/PDF |
| jakubkrehel/skills | Open source | 2026-07 | 11.0k stars | Polishing and review skills split by domain |
| anthropics/skills (frontend-design, etc.) | Big-company features | 2025-09 | 187.7k stars | The official general-purpose frontend design skill, not aimed at landing-page conversion |
| Motionsites-AI/claude-skill | Open source | 2026-10-26 | 1.7k stars | "Cinematic landing pages," with 50+ video assets built in, used to drive traffic to MotionSites |
| benjitaylor/skills(Stageshow) | Open source | 2026-10-26 | 2.1k stars | Fake product UI for marketing illustrations |

Compared with these, elaya is the thinnest in both commit count and engineering (16 commits, no linter, no rendering checks). In setproduct's stars-vs-commits scatter plot, it's the point furthest from the trend line. What it has that the others don't is two things: first, it does one thing only, landing pages, including asking about needs first and writing conversion copy; second, the author is a designer actually taking client work, and their portfolio is the demo.

### Breakdown of the growth tactics

1. **The author (confirmed) has long posted their work on X** : since August, nearly every work post links the repo, each post gets 7–80k views, and in two and a half months it reached 2.8k stars.

2. **A compilation video of the works went viral (confirmed)** : 10-24.9M views, directly bringing in 7k+ stars.

3. **Riding big tool names (confirmed)** : the copy name-drops @higgsfield, Claude, and Cursor. Higgsfield itself runs heavy paid ads everywhere, and posts like "I made this with Higgsfield…" get shared easily.

4. **KOLs and multilingual re-distribution (confirmed)** : Japanese @tetumemo, English @designertom and @neil_xbt, Medium, YouTube, IG reels, LinkedIn.

5. **GitHub trending #1 site-wide (confirmed)** : the repo was 60 days old, made trending, and 10-27/28 stayed on it for 3 days straight.

6. **Skills directories and SEO indexing (confirmed)** : skills.sh badges, ui-skills.com, typingmind, agentskillshub, skillleaderboard and other directories auto-indexed it.

7. **Timing (confirmed)** : "Designer skills week" plus the buzz about "AI landing pages all look the same" was hot.

8. No evidence of paid acquisition or star buying found; forks are normal, and there are no suspicious markers in the repo.

### Where the money comes from

- Open source itself earns nothing. The payoff comes through two channels: freelance client work at elaya.design, linked directly under the hit post; and SlopScan (slopscan.co), which audits live pages with the same rules and generates fix prompts, offered from 11-01 in the README banner. On TrustMRR, SlopScan made $14 in the last 30 days (https://trustmrr.com/startup/slopscan), so it has just started monetizing.

### How many people have seen it

- Already widely discussed: the original post with 877k views, big design-circle accounts like designertom, multiple comparisons and tutorials (setproduct, Medium, explainx, thakicloud, codearia), YouTube and IG remixes, and it hit #1 on GitHub's site-wide daily chart. SlopScan's revenue is public on TrustMRR, but only $14.

- The space is already crowded: several similar skills have 100k+ stars. Copycats have appeared, e.g. Motionsites-AI/claude-skill, created 10-26, "Free Claude skill: make Claude build award-level, cinematic landing pages", 1.7k stars in 11 days.

### Differentiation, and was it the reason it worked?

- On content: it only does landing pages, and puts "ask for requirements first, define structure, write conversion copy, apply a visual system" into one file. Editing one section swaps in your own style. That part is useful but not unique; taste-skill, hallmark, and impeccable cover most of the same ground.

- The real difference is the person and the demo: the author is a product designer with 8 years of experience, and has a full set of good-looking real work, released together with motion videos generated by Higgsfield. Viewers buy into "I can get this effect too"; how well the rules are written is secondary. So it took off mainly through distribution and demo material, with the product's uniqueness a secondary reason.

- Risks: 16 submissions, no validation tool; design-to-motion was added and 10-25 then "moved to its own repo", but we couldn't find where the new repo is (none in the author's public repos). After the burst, daily star growth has fallen back to around 10-27 to 150.

### What others can borrow

- Demo an open-source tool with finished work: a compilation video of "every piece I made with it these two months" beats any feature list.

- Use the open-source skill as a business card: link freelance booking and their paid tool (SlopScan) under the hit post, turning stars into client leads.

- Name-drop popular tools (Higgsfield/Claude/Cursor) to ride their traffic; keep all tunable values in one place so people can fork it, and more forks bring more exposure.

- Only run the big promotion after the repo is 60 days old, so it can catch the GitHub trending amplifier (this repo got it by coincidence).

- If you borrow someone else's work, credit them first, or you'll get publicly called out for credit (as with Leon Lin).

### Still unclear

- The view count of 7-29's launch post, and whether there were other sources before the August to October 2.8k stars (like Reddit or Product Hunt), were not checked.

- Which repo design-to-motion moved to; the GitHub API later rate-limited, so it wasn't checked again.

- The conversion rate from stars to actual installs: skills.sh install counts weren't retrieved.

- Actual revenue from SlopScan and client bookings.

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