convez.ai
SkillDesign & UIIniciante
moretothat

Grayscale

moretothat/grayscale

Converte imagens para escala de cinza — arquivo, pasta ou ZIP

Ver no GitHub
InicialPluginScore · de 0 a 100

Ferramenta leve que converte imagens para escala de cinza, com suporte a arquivos únicos, múltiplos, pastas e arquivos .zip. Funciona tanto como skill do Claude quanto como script CLI standalone, preservando transparência em PNG e outros formatos.

Instalação

$npx convez-ai moretothat/grayscale

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
0
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 3 semanas
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Converter lotes de fotos para preto e branco rapidamente
  • Processar arquivos ZIP completos mantendo estrutura de pastas
  • Automatizar workflow de pós-processamento de imagens
  • Integrar conversão de escala de cinza em pipelines de Design & UI

Veredito

Skill minimalista e prática que resolve um problema específico bem. Ideal para quem precisa de conversão batch sem overhead — zero configuração, funciona via Claude ou CLI.

README

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

grayscale

A tiny, dependency-light tool that converts images to grayscale. Point it at a single image, a bunch of images, a folder, or a .zip archive — it auto-detects what you gave it and converts everything in one pass.

It works two ways:

  1. As a Claude / Claude Code skill — drop it in your skills folder and just say "grayscale these" while uploading images or a zip.
  2. As a plain command-line script — python grayscale.py <anything>.

What it does

  • Converts any image(s) to grayscale.
  • Appends a -Grayscale suffix to every output file (Photo.jpg → Photo-Grayscale.jpg).
  • Preserves transparency for PNG / WebP / GIF / TIFF (grayscale + alpha); uses plain grayscale for JPEG / BMP.
  • Given a .zip, returns a matching Name - Grayscale.zip with every image converted, folder structure kept, and macOS __MACOSX/ junk + non-image files stripped out.
  • Handles a single file, many files, a directory, a zip, or any mix of those in a single command.

Install

git clone https://github.com/moretothat/grayscale.git
cd grayscale
pip install -r requirements.txt   # just Pillow

Usage (command line)

# A single image
python grayscale.py photo.jpg

# Several images at once
python grayscale.py a.png b.jpg c.webp

# A whole folder (add -r to include subfolders)
python grayscale.py ./my_photos
python grayscale.py ./my_photos -r

# A zip archive of images -> a zip of grayscale images
python grayscale.py album.zip

# Mix and match, and choose where output goes
python grayscale.py photo.jpg ./more_photos album.zip -o ./converted

Output defaults to ./grayscale-output/ (or /mnt/user-data/outputs/ when run inside the Claude/Cowork environment). Use -o <dir> to change it.

FlagMeaning
-o, --output-dirWhere to write results. Default: ./grayscale-output/.
-r, --recursiveWhen an input is a folder, also descend into subfolders.

Supported formats: PNG, JPG/JPEG, GIF, BMP, TIFF/TIF, WebP.

Use it as a Claude skill

This repo is the skill — SKILL.md plus the bundled grayscale.py.

  • Claude Code: copy this folder into your skills directory (e.g. ~/.claude/skills/grayscale/), then ask Claude to "grayscale" some images or a zip and it will run the script for you.
  • Claude.ai / Cowork: add it as a skill, then upload images or a zip and say "convert these to grayscale." The conversion runs automatically and the results (or a single output zip) come back for download.

The skill is intentionally zero-friction: it never asks follow-up questions, it just converts and hands back the files.

How it decides what to do

You give it…You get back…
One imageOne -Grayscale image
Several imagesEach converted individually
A folderEvery image inside, converted
A .zipA Name - Grayscale.zip with all images converted

Requirements

  • Python 3.9+
  • Pillow (pip install Pillow)

License

MIT

Ver o original no GitHub

Bom para

Relacionados