Content Producer Skill
narmaku/content-producer-skill
Gera vídeos MP4 animados e cinematizados com IA e ffmpeg
Ver no GitHubSkill que instrui Claude Code a gerar vídeos MP4 cinematizados e animados, frame-by-frame, com suporte a múltiplos tipos de conteúdo (demos, explainers, promos, tutoriais e showcases). Inclui tipografia cinética, ícones animados, transições suavizadas, música ambiente com duckingInteligencia de automático, narração de voz IA via Kokoro TTS com sincronização automática, paletas de cores brandizáveis e efeitos visuais (brilho ambiente, respiração, mockups de terminal).
Instalação
npx convez-ai narmaku/content-producer-skill1 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
- Criar vídeos de demonstração de produtos e features
- Produzir vídeos explicativos sobre conceitos técnicos
- Gerar promos, teasers e vídeos de anúncio de eventos
- Montar tutoriais em vídeo e guias passo-a-passo
- Compilar retrospectivas e reels de destaque
- Apresentações corporativas em formato vídeo
Veredito

Ferramenta versátil e bem-documentada para automação de produção de vídeo. Excelente cobertura de templates, personalizações e efeitos visuais; requer instalação de dependências (ffmpeg, Pillow, opcionais Kokoro e numpy), mas é plenamente open source e reutilizável.
README
A tradução para o português deste README ainda está na fila.
Content Producer Skill
A Claude Code skill for generating cinematic, animated MP4 videos — fully generic, brandable, and open source. Supports multiple content types, each with its own narrative structure, slide composition, and music style.
https://github.com/user-attachments/assets/88ecd7db-c93f-4d1c-9d2d-cc35cf5a6a30
What It Does
When triggered, this skill instructs Claude Code to generate a Python script that renders an MP4 video frame-by-frame using Pillow (PIL) and pipes raw RGB frames to ffmpeg for encoding. The resulting videos feature:
- 1920x1080 resolution at 30fps
- Configurable color palettes with multiple presets (blue, cyan, violet, amber, emerald, rose)
- Support for any TTF/OTF fonts (ships with Inter + JetBrains Mono defaults)
- Compatible with any open-source SVG icon set (Lucide, Tabler, Feather, etc.)
- Animated slides with kinetic typography, icon reveals, and eased transitions
- Crossfade transitions between slides
- Terminal mockups with typing animations
- Ambient glow effects with breathing animation
- Dynamic background music with 7 mood presets (upbeat, ambient, energetic, calm, cinematic, dark, lofi)
- Optional AI voice narration via Kokoro TTS (54 voices, 8 languages, voice cloning)
- Auto-ducking: background music volume reduces during narration
- Auto-duration: slide timing adjusts to match narration length
- Progress bar throughout the video
- Custom branding support (colors, fonts, logos)
Content Types
Choose the right template for your video:
| Type | Template | Best For | Duration | Music |
|---|---|---|---|---|
| Tech Demo | templates/demo.md | Product demos, feature walkthroughs | ~74s | Upbeat electronic (110 BPM) |
| Explainer | templates/explainer.md | Concepts, architectures, "what is X?" | ~71s | Ambient, thoughtful (90-100 BPM) |
| Promo | templates/promo.md | Events, launches, announcements | ~55s | High energy (120-130 BPM) |
| Tutorial | templates/tutorial.md | Step-by-step guides, how-to content | ~71s | Calm, focused (85-95 BPM) |
| Showcase | templates/showcase.md | Retrospectives, recaps, highlight reels | ~78s | Uplifting, cinematic (100-110 BPM) |
When It Triggers
Claude Code activates this skill when you ask to create a:
- Demo video, product video, or explainer video
- Animated presentation or motion graphics piece
- Event promo, teaser, or announcement video
- Tutorial or how-to video
- Showcase, retrospective, or highlight reel
- MP4 presentation of any content
Prerequisites
- Python 3 with
Pillowinstalled- Fedora:
sudo dnf install python3-pillow - Ubuntu:
sudo apt install python3-pil
- Fedora:
- ffmpeg installed on the system (uses
libopenh264codec)- Fedora:
sudo dnf install ffmpeg-free - Ubuntu:
sudo apt install ffmpeg
- Fedora:
- numpy (optional, for background music generation)
- Fedora:
sudo dnf install python3-numpy - Ubuntu:
sudo apt install python3-numpy
- Fedora:
- Kokoro TTS (optional, for AI voice narration)
pip install kokoro soundfile sudo dnf install espeak-ng # Fedora sudo apt install espeak-ng # Ubuntu - Fonts (at least one stack):
- Fedora:
sudo dnf install inter-fonts jetbrains-mono-fonts - Ubuntu:
sudo apt install fonts-inter fonts-jetbrains-mono
- Fedora:
- Icons — clone one of the recommended icon sets:
git clone https://github.com/lucide-icons/lucide.git
Skill Structure
video-skill/
SKILL.md # Main skill: rendering pipeline, animation, components, branding
README.md # This file
LICENSE # MIT License
templates/
demo.md # Tech demo / product video template
explainer.md # Explainer / educational video template
promo.md # Event / product promo template
tutorial.md # Step-by-step tutorial template
showcase.md # Showcase / retrospective template
references/
design-system.md # Color palettes, typography, spacing rules
story-arcs.md # Narrative structures for storytelling
icons.md # Open-source icon libraries guide
music.md # Dynamic music generation system (7 mood presets)
media-assets.md # Embedding images and screen recordings
branding-guide.md # How to set up branded video production
How It Works
- Claude reads the shared references (design system, story arcs, icons, media assets)
- If the user provides assets (images, screen recordings), Claude examines them to understand their content
- Based on your topic, it selects the appropriate content type template
- Reads the template for slide structure, design guidance, and music style
- Auto-maps provided assets to the most appropriate slides and writes matching narration
- Optionally applies your custom brand colors, fonts, and logo
- Generates a self-contained Python script (
make_video.py) that:- Defines slide functions, each returning a PIL Image for a given frame
- Renders all frames sequentially with crossfade transitions
- Pipes raw RGB bytes to ffmpeg via stdin
- Optionally generates and muxes background music
- Optionally generates TTS narration per slide (auto-adjusting durations)
Usage
Create a demo video about how our CLI tool simplifies database migrations
Create an explainer video about what WebAssembly is and why it matters
Create a promo video for "Open Source Summit 2025" happening October 10-12 in Dublin
Create a tutorial video on setting up SSH key authentication
Create a showcase video recapping our open source project's first year
With custom branding:
Create a demo video about our monitoring tool using amber accent colors and Outfit font
With narration:
Create a narrated tutorial video about setting up Docker on Linux
With user-provided assets:
Create a demo video about our new dashboard. I have these assets:
- assets/install-recording.mp4 (screen recording of the setup process)
- assets/dashboard-overview.png (screenshot of the main UI)
- assets/metrics-page.png (screenshot of the metrics panel)
Custom Branding
Produce on-brand videos by creating a brand/ directory with your organization's assets:
brand/
brand.md # Brand config: colors, fonts, rules, voice
logo.png # Your logo (transparent PNG)
fonts/ # Custom font files (optional)
icons/ # Custom icon set (optional)
Just create brand/brand.md with your colors and logo, and every generated video
will automatically use your brand. Example:
Create a demo video about our new API gateway
Claude reads brand/brand.md, applies your brand colors, fonts, and logo — no
extra flags or configuration needed.
The skill also works alongside existing brand skills. If your organization already has
brand reference files (color palettes, typography guides), you can copy or reference
them from your brand/ directory.
See references/branding-guide.md for the full setup guide, configuration reference,
and examples.
License
MIT License. See LICENSE for details.
Acknowledgments
This skill builds upon the work of:
- Brian Smith — red-hat-video-skill, the original video generation skill whose rendering pipeline, animation system, and visual component patterns served as the foundation for this project.
- Todd Wardzinski — red-hat-quick-deck, whose story arc structures, slide design principles, and design system patterns were adapted and extended here.
Both projects are MIT licensed. The brand-specific elements were replaced with a generic, configurable design system to make this skill usable by anyone.