PromptShot
kcperez/promptshot
Screenshots de página inteira direto para seu agent de código
Ver no GitHubExtensão para navegador que captura a página inteira em um clique e copia o caminho do arquivo na área de transferência, pronto para colar em agentes de IA como Claude Code, Codex e Cursor. Automatiza o fluxo de trabalho entre browsers e assistentes de código, sem armazenar dados em servidores — tudo fica local na máquina do usuário.
Instalação
npx convez-ai kcperez/promptshot1 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
- Capturar interfaces visuais e enviar para Claude Code analisar layout
- Debugar páginas web compartilhando screenshots com agentes de IA
- Acelerar desenvolvimento passando capturas direto do navegador para Cursor
Veredito

Ferramenta focada e bem executada que reduz atrito entre navegação e coding agents. Priva, sem overhead — resolve um problema real do fluxo dev moderno com IA sem complicações.
README
A tradução para o português deste README ainda está na fila.
PromptShot
Full-page screenshots, straight into your coding agent.
One click captures the entire page, saves it to a single folder, and drops the file's path onto your clipboard. Paste it into Claude Code, Codex, or Cursor and your agent reads the screenshot instantly. No more digging through Downloads.
- 🖱️ One click — full-page scroll-and-stitch capture, or
⌥⇧S - 📋 Path on your clipboard — exactly what a terminal agent needs (
@/path/shot.png) - 🖼️ Image mode — copy the PNG itself for web chat boxes (Cursor, claude.ai)
- 🧹 No graveyard — shots land in one
agent-shots/folder and old ones auto-delete - 🔒 Private — no servers, no account, no analytics, no host permissions
Privacy
PromptShot collects nothing. Screenshots are saved to your own machine and never
transmitted. It requests only activeTab, scripting, downloads, and
storage, and has no access to the content of the pages you visit.
How it works
- Clicking the toolbar icon opens a small popup. The popup asks the background service worker to capture and save, then copies the path itself — a popup is a focused context, which is the only reliable place to write the clipboard from an MV3 extension (a toolbar click blurs the page, and offscreen documents are never focused).
- Full-page capture scrolls the tab in viewport-sized steps, captures each with
chrome.tabs.captureVisibleTab, and stitches them onto anOffscreenCanvas. Fixed/sticky elements are hidden after the first frame so they don't repeat. - The keyboard shortcut keeps page focus, so it copies in-page directly.
Develop
Built with WXT. Requires Node 18+ and pnpm.
pnpm install
pnpm dev # launches Chrome with hot-reload
pnpm build # outputs .output/chrome-mv3
Icons are generated from scripts/make-icons.mjs (node scripts/make-icons.mjs).
Load it unpacked
pnpm build- Open
chrome://extensions, enable Developer mode - Load unpacked → select
.output/chrome-mv3
Works on any Chromium browser (Chrome, Edge, Brave, Arc).
License
MIT © 2026 Kevin Castro