IMG for AI

You paste a screenshot into your AI coding agent — an error, a UI, a diagram — and a Retina capture quietly spends thousands of vision tokens plus a chunk of the context window you'd rather keep for code. IMG for AI right-sizes it in one drag. Drop the image on the tray and it's instantly scaled to the resolution the model actually uses, so it costs a fraction of the tokens — then paste it straight into your agent. Why it works: vision models count tokens by image resolution (pixels), not file size. A full-res screenshot carries far more pixels than the model needs; scaling it down cuts the tokens, and the context it consumes, accordingly. Most useful when you send large screenshots to AI coding tools and agents. Works great with the likes of Claude Code and Codex. • One-drag optimize from anywhere — the tray appears the moment you start dragging • Keep your context window for code, not oversized images • Resolution presets tuned for AI vision; optional sharpening keeps text and UI readable • PNG or WebP — smaller uploads, too • Lives in your menu bar — lightweight, never in the way • 100% on-device — no account, no network, no data collected