Claude Code Skills
tylerprogramming/claude-skills
Workflows automatizados em slash commands para produtividade e criação de conteúdo
Ver no GitHubColeção de skills reutilizáveis para Claude Code que automatizam workflows via comandos `/` direto do terminal. Inclui automações para fitness, criação de conteúdo (YouTube, TikTok, Reels), análise de dados, geração de documentos e processamento de mídia.
Instalação
npx convez-ai tylerprogramming/claude-skills1 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
Casos de uso
- Logging de treinos e nutrição com sync a Supabase
- Pesquisa e geração de conteúdo para YouTube, TikTok e Reels
- Análise de vídeos, transcrição e otimização de títulos/thumbnails
- Geração de documentos (PRDs, Custom GPTs, planos de refeição)
- Processamento em batch de imagens e remoção de backgrounds
- Gerenciamento de comentários e upload de vídeos em plataformas
Veredito

Coleção bem estruturada e prática de skills Python para potencializar Claude Code. Cobre casos reais desde fitness tracking até content creation multi-plataforma, mas requer conhecimento técnico para setup de APIs/Supabase e manutenção.
README
A tradução para o português deste README ainda está na fila.
Claude Code Skills
My personal collection of Claude Code skills — reusable slash commands that automate workflows directly from the terminal.
What are Skills?
Skills are saved workflows for Claude Code. Instead of explaining what you want every time, you write it down once as a SKILL.md file, and Claude remembers how to do it. Invoke them with /skill-name in Claude Code.
Skills
| Skill | Command | Description |
|---|---|---|
| Journal | /journal | Daily standup logger with weekly summaries |
| OPM | /opm | One Punch Man workout logger — logs the 100/100/100 circuit, runs, and water to the Supabase fitness-tracker project, with Saitama-style coaching |
| Nutrition | /nutrition | Food logging, eating review vs targets, meal suggestions, and recipe cards on the Supabase fitness-tracker project |
| OPM Graph | /opm-graph | Live training dashboard (self-contained HTML) reading charts directly from the Supabase fitness-tracker project |
| OPM Review | /opm-review | Weekly training + nutrition review written to the weekly_review table |
| Fitness | /fitness | (Legacy SQLite version — superseded by /opm + /nutrition.) Track workouts and nutrition in a GitHub-style contribution grid. Powered by a full React + Hono + SQLite app (included in fitness/app/). |
| Meal Plan | /meal-plan | (Legacy — superseded by /nutrition.) Generate a weekly meal plan with shopping list from your recipe folder |
| PRD | /prd | Generate a Product Requirements Document for a new feature |
| Quiz | /quiz | Interactive quiz/coaching from any document — mock interviews, rapid fire, deep practice, study review |
| Ralph | /ralph | Convert PRDs to prd.json format for the Ralph autonomous agent system |
| Resize | /resize | Batch resize images to social media presets |
| Remove BG | /rmbg | Remove backgrounds from images to transparent PNGs |
| Save Idea | /yt-save-idea | Save a YouTube video idea to your ideas tracker |
| Thumbnail | /thumbnail | Generate YouTube thumbnails via Kie.ai (Nano Banana 2, Nano Banana Pro, Seedream 4.5) |
| TikTok | /tiktok | Research TikTok trends by hashtag via Apify — scrape, analyze, and suggest content ideas |
| Transcribe | /transcribe | Transcribe YouTube videos or local audio/video files using OpenAI Whisper |
| YT | /yt-package | Full YouTube video package — analysis, research, titles, hooks, script, description, filming guide |
| YT Search | /yt-search | Search YouTube by keywords via yt-dlp, sort by views, generate report |
| Shorts | /shorts | Weekly short-form content package: 5 short scripts (YT Shorts/TikTok/Reels) + 2 Instagram carousel outlines from yt-search research |
| Custom GPT | /custom-gpt | Build a complete Custom GPT package (system-prompt, KB files, README) from course transcripts or raw text |
| Analytics | /yt-analytics | YouTube channel analytics - views, retention, traffic sources, per-video breakdown, title/thumbnail A/B tracking |
| YouTube | /yt-upload | Upload videos to YouTube and manage existing videos via the YouTube Data API v3 (titles, descriptions, tags, thumbnails, scheduling, comments) |
| TikTok Replier | /tiktok-replier | Two-stage TikTok comment management: Apify reads comments, Playwright posts replies using a persistent logged-in browser profile |
AI Media Generation — Kie.ai
One skill per model, all sharing _kie/kie_client.py (need KIE_API_KEY in ~/.claude/.env). See _kie/MODELS.md for the full model id + schema reference.
| Skill | Command | Description |
|---|---|---|
| Kie Seedance Video | /kie-seedance-video | ByteDance Seedance text-to-video + image-to-video (Seedance 2 / 2-Fast / 1.5 Pro / 1.0 Pro·Lite, optional audio) |
| Kie Seedance Image | /kie-seedance-image | ByteDance Seedream text-to-image + multi-image edit (2K/4K) |
| Kie Kling Video | /kie-kling-video | Kling (Kuaishou) text/image-to-video (2.6 / 3.0 / 2.1 Master·Pro, native sound, multi-shot) |
| Kie Veo Video | /kie-veo-video | Google Veo 3 / 3.1 video — native audio, up to 4K (Quality/Fast/Lite). Uses Kie's dedicated /veo endpoint |
| Kie Wan Video | /kie-wan-video | Alibaba Wan text/image-to-video (2.7 / 2.6 / 2.5, first+last frame, audio drive) |
| Kie Nano Banana | /kie-nano-banana | Google Nano Banana image gen/edit — all 3 variants (Nano Banana, 2, Pro), up to 4K |
AI Media Generation — Higgsfield
Higgsfield uses the official first-party skills (not vendored in this repo). Install them once with:
npx skills add higgsfield-ai/skills
They land in ~/.agents/skills/ (symlinked to Claude Code) and authenticate via the Higgsfield CLI — run higgsfield auth login once (needs a Higgsfield account with credits). No API key in .env.
| Skill | Command | Description |
|---|---|---|
| Higgsfield Generate | /higgsfield-generate | Image/video across 30+ models (Soul V2/Cinema, Seedance 2.0, Kling 3.0, Nano Banana, GPT Image 2), image-to-image, image-to-video, Marketing Studio ads, and the Virality Predictor |
| Higgsfield Soul ID | /higgsfield-soul-id | Train a Soul Character (face/identity model) from a few photos; reuse via --soul-id for consistent faces |
| Higgsfield Product Photoshoot | /higgsfield-product-photoshoot | Brand-quality product imagery (studio, lifestyle, hero/banner, ad creative, virtual try-on…) |
| Higgsfield Marketplace Cards | /higgsfield-marketplace-cards | Marketplace listing images — main + secondary product shots and A+ content modules |
Installation
These are global skills. To use them, copy the skill folders to your Claude Code skills directory:
# Clone the repo
git clone https://github.com/tylerprogramming/claude-skills.git
# Copy to your Claude skills folder
cp -r claude-skills/* ~/.claude/skills/
Fitness App Setup
The /fitness skill includes a full-stack app. After installing the skills:
cp -r ~/.claude/skills/fitness/app ~/fitness-app
cd ~/fitness-app
bun install # or: npm install
bun run dev # starts both React frontend and Hono API server
- Frontend: http://localhost:5173
- API: http://localhost:3001
- SQLite DB is auto-created at
~/fitness-app/fitness.dbon first run
Skill Structure
Each skill has:
SKILL.md— Instructions for Claude (name, description, flow)- Optional scripts (
.py,.sh) for complex operations
Secrets & Per-User Setup
None of the secrets in this repo are committed. After cloning, you'll need to create your own:
-
.env— copy.env.exampleto.envand fill in your own keys (OpenAI, Kie.ai, Apify, Resend, Gmail, Skool). -
/yt-upload— uses the YouTube Data API v3:- Create OAuth credentials at https://console.cloud.google.com/apis/credentials (Desktop app) and save the downloaded file to
~/credentials.json. - First run:
python ~/.claude/skills/yt-upload/yt.py authto walk the OAuth flow. This writesyoutube/token.json(gitignored).
- Create OAuth credentials at https://console.cloud.google.com/apis/credentials (Desktop app) and save the downloaded file to
-
/tiktok-replier— uses a persistent Chromium profile attiktok-replier/data/profile/to stay logged into TikTok. The first run will open a browser; log in once and the session is reused. Thedata/directory (profile, queues, logs, scraped HTML) is gitignored. -
/yt-analytics— needs its own OAuth token atanalytics/yt_token.json(gitignored).
The following are also intentionally gitignored: *.zip backups, __pycache__/, .venv/, and the symlinks pointing to local-only agent skills (hyperframes*, gsap, find-skills, claude-design-hyperframes).
License
MIT — use these however you want.