convez.ai
SkillUtilidadesIntermediário
tekvisions

Skill Index

tekvisions/skill-index

Diretório de skills Claude ranqueado por momentum, atualizado diariamente

Ver no GitHub
InicialPluginScore · de 0 a 100

Diretório vivo e atualizado diariamente de skills para Claude Code e AI agents, ranqueados por momentum usando sinais reais do GitHub. Indexa mais de 273 repositórios em 9 categorias, com recomputação automática a cada 24 horas e sem viés editorial ou patrocínio.

Instalação

$npx convez-ai tekvisions/skill-index

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 1 semana
Versão
—
Issues abertas
0
Contribuidores
3

Casos de uso

  • Descobrir skills e subagents populares e em crescimento no ecossistema Claude
  • Pesquisar e filtrar ferramentas por categoria, momentum, recência ou número de stars
  • Monitorar tendências do ecossistema de IA agents através de rankings imparciais

Veredito

Ferramenta sólida de curação que resolve um problema real: encontrar skills de qualidade em um ecossistema fragmentado. O modelo de momentum (55% stars + 32% recência + 13% crescimento) privilegia projetos ativos sobre legados gigantes, e a atualização automática mantém relevância. Recomendado para quem quer se manter atualizado no ecossistema Claude sem ruído.

README

A tradução para o português deste README ainda está na fila.

The Skill Index

Every Claude Code & AI-agent skill, ranked by momentum.

Live   Updated   Indexed   Built by

Open the live index →


What it is

The Skill Index is a living, self-updating directory of Claude Code skills, subagents, slash-commands, hooks, and the curated collections that hold them.

Every repository is ranked by momentum — not by total stars, and not by who paid for placement. The whole index is recomputed every day from live GitHub signals and republished automatically. No editors, no vendor decks, just the signal.

🔗 Live: skill.kymatalabs.com · part of The Living Indexes

Features

  • 📊 Momentum ranking — a 0–100 score blending stars, push-recency, and how fast a repo is rising
  • 🔄 Self-updating — a daily GitHub Action recomputes and redeploys; the index is never stale
  • 🗂 9 categories — Collections · Commands & Hooks · Subagents · MCP & Tools · General Skills · Design & UI · Memory & Context · Quality & Review · Writing & Content
  • 🔎 Instant search, filter & sort — by momentum, stars, or newest, fully client-side
  • 📄 An SEO landing page for every entry — title, meta, canonical, Open Graph, and SoftwareSourceCode JSON-LD
  • 🌗 Light & dark themes, RSS feed, and an llms.txt for AI crawlers

How it works

build_data.py   →   data.json   →   gen_details.py   →   deploy.py   →   ⟳ daily cron
 (GitHub API)       (the index)     (per-entry pages)    (Vercel REST)    (GitHub Action)
  1. build_data.py searches GitHub across several queries, de-duplicates, filters to real skills (precision over recall — a directory you can trust beats a noisy dump), categorizes, and scores each by momentum → writes data.json + the SEO surfaces (sitemap.xml, rss.xml, robots.txt, llms.txt).
  2. gen_details.py renders one fully-SEO'd landing page per entry under p/<slug>/.
  3. gen_og.py renders the Open Graph share card.
  4. deploy.py ships the static site to Vercel via the REST API.
  5. A daily GitHub Action runs the whole pipeline and redeploys — so the index reflects the ecosystem as it is today.

How momentum is scored

momentum = 0.55 · log-scaled stars   (popularity, compressed so giants don't flatten everything)
         + 0.32 · push recency       (full credit if pushed today, decaying to zero by ~180 days)
         + 0.13 · rising-newness     (a bonus for young repos gaining stars fast)

A repo that shipped a release this week outranks a bigger repo that's gone quiet — momentum, not legacy.

Run it locally

GITHUB_TOKEN=ghp_xxx  python3 build_data.py   # recompute the index from live GitHub signals
python3 gen_details.py && python3 gen_og.py    # render the per-entry pages + OG card
python3 -m http.server 8080                    # open http://localhost:8080

No build step, no framework — static HTML/CSS/JS by design, so it loads instantly and hosts anywhere.

Tech

Python (data pipeline) · static HTML/CSS/JS (no framework) · Vercel (hosting) · GitHub Actions (daily cron) · the GitHub Search API as the single source of truth.

Part of The Living Indexes

This is one of a fleet of self-updating indexes mapping the AI-builder ecosystem:

→ See them all on indexes.kymatalabs.com


© 2026 Kymata Labs LLC. Skill Index is a self-updating reference, recomputed daily.

Ver o original no GitHub

Bom para

Relacionados