๐ Weekly Skill Leaderboard โ October 16, 2026
๐ Weekly Skill Leaderboard โ October 16, 2026
Compiled Friday, October 16, 2026 ยท 6 categories ยท 18 skills ranked ยท Data: skills.sh, Linkly AI (updated 2026-10-15), ClawHub, GitHub
This was the week Microsoft walked into the skills game, a wave of "superpowers"-branded media packs took over the 24-hour trending board, and ClawHub registry skills from the OpenClaw ecosystem crossed over into mainstream leaderboards. Meanwhile, our own 30-day dedup window hit a record: 82 skills entered the cooldown โ which means every podium below is the freshest lineup we've published in a month. One champion returns: Ponytail defends its #1 with a rebuilt v5 and receipts to prove it.
๐ Category Rankings
๐งโ๐ป Coding & Development
| Rank | Skill | Developer | Stars / Installs | Format | Pricing |
|---|---|---|---|---|---|
| ๐ฅ | Ponytail (v5) | DietrichGebert | 163.9kโ / 8.8k forks | SKILL.md + AGENTS.md + plugin hooks | Free, MIT |
| ๐ฅ | improve-codebase-architecture | Matt Pocock | 1.1M installs (skills.sh #5) | SKILL.md | Free, MIT |
| ๐ฅ | domain-modeling | Matt Pocock | 820,976 installs ยท WIS 62.9 | SKILL.md | Free, MIT |
Why these ranked here โ Ponytail is our reigning #1 and it defended the crown the right way: Ponytail 5 is a ground-up rebuild benchmarked on 39 agentic tasks against a real FastAPI + React repo (Opus 5.5, 5 runs each, method published Oct 7): -53% code, -41% time, -26% cost, -45% tokens versus the no-skill baseline โ while raising quality (98% of risky logic ships with a test vs 68% without; its /ponytail-review finds 100% of planted problems vs 87%). Still one prompt (skills/ponytail/SKILL.md), still 20 agents, still free. improve-codebase-architecture is the quiet workhorse of Matt Pocock's 292kโ
collection โ it surveys a codebase for module-deepening opportunities, hands you a visual HTML report, then grills you through the fix. At 1.1M installs it's the #5 skill on all of skills.sh. domain-modeling rounds out the podium by fixing the root cause of agent verbosity: it challenges your project's terms against a shared glossary, stress-tests edge cases, and updates GLOSSARY.md and ADRs inline โ 820,976 installs and climbing.
โ๏ธ Cloud & DevOps โ ๐ new category
| Rank | Skill | Developer | Installs / WIS | Format | Pricing |
|---|---|---|---|---|---|
| ๐ฅ | microsoft-foundry | Microsoft | 565,549 ยท WIS 62.8 | SKILL.md | Free |
| ๐ฅ | azure-diagnostics | Microsoft | 557,331 ยท WIS 62.8 | SKILL.md | Free |
| ๐ฅ | azure-ai | Microsoft | 556,410 ยท WIS 62.4 | SKILL.md | Free |
Why these ranked here โ Microsoft didn't ship one skill; it shipped an army. The azure-skills family put roughly twenty azure-* entries into Linkly's top-40 within their first weeks โ every one of them sitting between 503k and 566k installs. microsoft-foundry leads on Worth-Installing Score (62.8, tied with azure-diagnostics), teaching agents Foundry-side resource management via the Skills REST API; azure-diagnostics is the troubleshooting workhorse; azure-ai covers model deployment and AI resource plumbing. Enterprise cloud ops just became a skill-native discipline, and Microsoft owns the category on day one.
๐จ Design & Creative
| Rank | Skill | Developer | Installs / WIS | Format | Pricing |
|---|---|---|---|---|---|
| ๐ฅ | web-design-guidelines | Vercel Labs | 736,963 ยท WIS 65.4 | SKILL.md | Free, MIT |
| ๐ฅ | design-mobile-apps | designed-by-ai (Sleek) | 837.9K ยท 21.9K in 24h | SKILL.md + OpenAPI | Free (Sleek API) |
| ๐ฅ | uizze-ui | uizze | 19.5K installs in 24h (#14 trending) | SKILL.md | Free |
Why these ranked here โ web-design-guidelines is the design-review skill with actual teeth: 100+ rules spanning accessibility (aria, semantic HTML, keyboard), focus states, forms, animation (prefers-reduced-motion, compositor-friendly transforms), typography (curly quotes, tabular-nums), images, performance, dark mode, touch and i18n โ from Vercel's 32.5kโ
official collection. design-mobile-apps is the sleeper hit: 837.9K installs and still pulling 21.9K/24h, it drives mobile screen design through the Sleek Design API and implements the results in React Native or HTML โ and it passed the Gen Agent Trust Hub security audit. uizze-ui takes bronze on pure momentum: #14 on the 24-hour trending board. Transparency note: it comes from the same vendor ecosystem as two skills we reviewed on Oct 9 (ios-design, ui-taste from uizze.sh), but it's a different package (uizze/uizze) that has never been individually reviewed โ included on merit, flagged for the record.
๐ฌ Media Generation
| Rank | Skill | Developer | Installs | Format | Pricing |
|---|---|---|---|---|---|
| ๐ฅ | nano-banana-pro | steipete (Peter Steinberger) | 266K combined (ClawHub 107,520 + SkillHub CN 158,652) ยท WIS 63.1 | SKILL.md (CLIPrompt) | Free (API costs) |
| ๐ฅ | ai-image-generation | magentosh | 30.4K in 24h (#2 trending) ยท pack 121.7K | SKILL.md pack | Free |
| ๐ฅ | image-to-video | gencraft-labs | 15.0K in 24h (#21 trending) ยท pack 58.9K | SKILL.md pack | Free |
Why these ranked here โ nano-banana-pro is the cross-registry champion: Peter Steinberger's skill wraps Google's Nano Banana Pro (Gemini 3 Pro Image) for generate-and-edit workflows โ text-to-image and image-to-image at 1K/2K/4K with --input-image editing โ and it's one of the rare skills ranked highly across two registries simultaneously (ClawHub and Tencent SkillHub CN), which is exactly the multi-ecosystem validation our WIS score rewards. Behind it, a genuinely new trend: superpowers-branded media packs. magentosh's ai-image-generation hit #2 on skills.sh trending within 24 hours (30.4K installs, 121.7K across its pack), and gencraft-labs' image-to-video anchors the video side (15.0K/24h, 58.9K pack total). Raw velocity, brand-new entrants, zero prior coverage.
๐ฃ Harness & Orchestration
| Rank | Skill | Developer | Installs | Format | Pricing |
|---|---|---|---|---|---|
| ๐ฅ | self-improving | ivangdavila | 726K combined (ClawHub 211,434 + SkillHub CN 514,815) ยท WIS 61.8 ยท v1.2.16 | SKILL.md + memory files | Free |
| ๐ฅ | reddit-automation | flowkit-labs | 940.4K all-time (#13) ยท 14.9K/24h | SKILL.md | Free |
| ๐ฅ | twitter-automation | its-a-skill-issue | 10.2K in 24h (#40 trending) | SKILL.md pack | Free |
Why these ranked here โ self-improving is the most interesting idea on this week's board: ivangdavila's skill gives agents a tiered memory system โ episodic memory written to JSON per task, pattern-confidence scoring that decays unused patterns (pruned below 0.3 confidence), and permanent learning from user corrections and failed operations. At v1.2.16 with 726K combined installs and its own third-party security report, it's the OpenClaw ecosystem's flagship export. reddit-automation is the scale story โ 940.4K all-time installs makes it one of the biggest workflow-orchestration skills in existence, still moving 14.9K/day. twitter-automation takes the last podium spot on the same "superpowers" pack wave, proving the brand-clone formula travels across content platforms.
๐งช Testing & Eval
| Rank | Skill | Developer | Installs | Format | Pricing |
|---|---|---|---|---|---|
| ๐ฅ | tdd | Matt Pocock | 1,084,469 ยท WIS 60.9 | SKILL.md | Free, MIT |
| ๐ฅ | triage | Matt Pocock | 944.7K (skills.sh #12) | SKILL.md | Free, MIT |
| ๐ฅ | diagnosing-bugs | Matt Pocock | 784.4K (skills.sh #26) | SKILL.md | Free, MIT |
Why these ranked here โ Matt Pocock's engineering collection (292.0kโ
, 545 commits) didn't win one podium spot in this category โ it swept it, and the sweep is earned: tdd (1,084,469 installs) enforces the red-green-refactor loop with explicit guidance on what makes good vs bad tests; triage (944.7K) moves issues through a state machine of triage roles with label-driven gates โ the quality-gatekeeping layer most agent workflows skip; diagnosing-bugs (784.4K) wraps debugging in a disciplined gated loop: build a feedback loop that goes red โ minimize โ hypothesize โ instrument โ fix โ regression-test. Three skills, one philosophy: fundamentals over vibes. None had been individually reviewed until now โ a year's worth of daily-use credibility, finally on the board.
๐ Movers & Shakers
- ๐ Biggest riser: the "superpowers" pack wave โ magentosh (121.7K pack installs, #2 trending), 101-skills (41.1K), bankai-skills (40.7K), its-a-skill-issue all charting simultaneously in 24 hours. Media-generation packs are minting installs at record speed.
- ๐ New entry of the week: Microsoft's azure-skills family โ ~20 skills flooding Linkly's top-40 in their first weeks. The category "Cloud & DevOps" exists on this leaderboard because of them.
- ๐ Cooled down: the grill family (
grill-me1.4M,grill-with-docs1.2M installs โ still #2/#3 all-time on skills.sh) hit the 30-day dedup wall after two reviews inside a month. No momentum problem โ purely editorial cooldown. They'll be eligible again in November. - โ๏ธ Also notable:
frontend-design(WIS 77.1) andagent-browser(WIS 80.8) remain the top-2 skills on Linkly overall โ but both were ranked here in the last two weeks, so they cede their podiums this edition.
๐ฎ Next Week's Watch List
- OmniRoute (diegosouzapw) โ MIT-licensed local AI gateway: 231โ352 providers behind one OpenAI-compatible endpoint, RTK + Caveman token compression claiming 15โ95% savings, 17 routing strategies. Gateway-side rather than SKILL.md-native today; if it ships a proper skill package, it owns a future "LLM Gateway & Routing" category.
- hyperframes (heygen-com) โ the core skill sits at 986.0K all-time installs, but sibling skills are in cooldown until early November. It re-qualifies with the full HeyGen stack behind it.
- The pack-authenticity question โ with four "superpowers"-branded packs charting in one day, someone needs to ask which ones are sanctioned, which are clones, and whether registry branding rules exist at all. We're on it.
Methodology
Ranking criteria, in priority order: community signal (stars/installs/trending velocity) โ recency (active maintenance) โ modularity (clean SKILL.md, structured scripts, docs) โ real-world utility โ innovation. 30-day dedup enforced against our reviewed-skills database (82 skills excluded this week); the only exception is a defending champion with a major version update (Ponytail v5).
Sources: skills.sh ยท skills.sh trending ยท Linkly AI rankings (2026-10-15) ยท GitHub: DietrichGebert/ponytail ยท GitHub: mattpocock/skills ยท GitHub: microsoft/azure-skills ยท GitHub: vercel-labs/agent-skills ยท GitHub: remotion-dev/skills ยท ClawHub: nano-banana-pro ยท ClawHub: self-improving ยท skills.sh: design-mobile-apps
Ranked by AgentSkillReview. Data sourced from GitHub, skills.sh, Linkly AI, ClawHub and community forums. Skills only โ no platforms, no SaaS, no IDEs.