convez.ai
Skill
gooseworks-ai

goose-video

gooseworks-ai/goose-video

Ver no GitHub
InicialPluginScore · de 0 a 100

Video generation skills for AI agents like Claude Code, Cursor, and Codex. Orchestrates Google Veo 3, Seedance, Higgsfield, ElevenLabs and other AI models to make short-form videos

Instalação

$npx convez-ai gooseworks-ai/goose-video

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
19
Forks
2
Acompanhando
0
Saúde
37%
Último commit
há 5 meses
Versão
—
Issues abertas
2
Contribuidores
3

README

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

goose-video

Open agent skills for AI video production.

Drop a skill folder into any agent that reads SKILL.md (Claude Code, Cursor, Goose, Codex CLI, …), then describe the video you want to make. The agent walks you through it.

Skills

SkillWhat it makesValidated on
animated-explainer-ad9:16 ~30–45s absurdist Pixar 3D explainer ad where a personified villain narrates their own defeat by the product. Bright cartoon characters, single-narrator VO, real-product end card, burned captions, whimsical Pixar/Disney score. ~$15–25 per cut.HUM Nutrition "Big Chill" · Soteri Skin "Eczema, the pH villain"
imessage-video-ad9:16 ~17–22s iMessage conversation reveal ad — bubbles pop in over time, composer types char-by-char, real Apple iMessage SFX on every send/receive, lofi bed, framed iPhone on a flat-lay desk, brand end card with a promo code. One continuous Playwright recording. ~$0.20–0.65 per cut.GooseWorks internal builds (product-flex + result-flex)

More coming. If you ship something using one of these, would love to see it.

How a skill is structured

Each skill is a self-contained folder:

<skill>/
├── SKILL.md          ← agent reads this to drive the workflow
├── README.md         ← human-facing intro
├── .env.example      ← API keys needed
├── requirements.txt  ← Python deps
├── scripts/          ← deterministic glue (FFmpeg + provider SDKs)
├── examples/         ← ready-to-paste kickoff prompts
└── references/       ← worked case studies

The agent reads SKILL.md to know:

  • when to invoke this skill (trigger phrases, use cases)
  • what inputs to gather from the user
  • how to walk the production pipeline phase-by-phase
  • when to pause for human review
  • what scripts to call and with what arguments

Scripts are kept small, single-purpose, and provider-grounded. No giant config schemas — the agent stitches them.

Quickstart

Paste this into Claude Code (or any agent that can run shell + read files):

Clone https://github.com/gooseworks-ai/goose-video into ./goose-video, then
read the skill at ./goose-video/skills/templates/animated-explainer-ad/SKILL.md
and walk me through it to make an animated explainer ad.

Before starting:
  1. cd into ./goose-video/skills/templates/animated-explainer-ad
  2. pip install -r requirements.txt
  3. Help me create a .env from .env.example — I'll need a FAL_KEY (https://fal.ai)
     and ELEVENLABS_API_KEY (https://elevenlabs.io).

Then ask me for: the brand, the product, the problem we're personifying as the
villain, the ownable mechanism, and the audience. From there, follow the SKILL.md
8-phase pipeline (intake → anchors → VO → keyframes → clips → end card → music →
compose) and pause at each human review gate so I can approve before paying for
the next phase.

See each skill's own README for full setup details and additional example prompts.

Built by

GooseWorks — AI agents that make videos and run ads. We use these skills internally for client work; publishing them so others can build on the same patterns.

License

MIT — use freely, attribution appreciated.

Ver o original no GitHub

Bom para

Relacionados