Open Source Chrome Extension

Stop describing
what you see

A screenshot with an arrow beats a paragraph of explanation. Capture, annotate, and share visual context with your AI coding tools.

Add to Chrome Setup Guide
npm install -g @snip-browser/ai-companion
The Problem

Your AI can't see your screen

AI coding tools are powerful, but they're blind. When something looks wrong, you're stuck typing what you see instead of showing it.

Without Snip

You describe the issue in words, hope the AI understands, go back and forth clarifying.

"The button on the right side of the
search bar has the wrong icon. It's a
sparkle thing but should be a wizard.
No, not that one, the AI Mode button."

With Snip

Capture the screen, draw an arrow pointing to the problem, and your AI sees exactly what you mean.

snip-wait --url "https://google.com" \
  --reason "change this icon"

# AI reads the annotated screenshot
# and knows exactly what to fix
How It Works

Three steps to visual context

Your AI requests a visual, you capture and annotate, and it sees what you see.

1

AI Requests

Your AI tool runs snip-wait to open a URL and request a visual capture. Chrome opens with the Snip area selector ready.

2

You Capture

Select a region, annotate with arrows, text, highlights, tags, or blur sensitive info. Click Copy when done.

3

AI Sees

The annotated screenshot is saved locally. Your AI reads the image and responds with full visual context.

Annotation Tools

Simple, powerful tools

Fast, focused tools for annotating screenshots. Each one has a keyboard shortcut.

Rectangle

Outline, highlight, or blur regions. Three modes for different needs.

R

Text

Add labels, callouts, or notes directly on the image.

T

Arrow

Point to exactly what you mean. No more "the thing on the right."

A

Tag

Labeled callouts with leader lines. Color-coded for organizing feedback.

G

Blur

Pixelate sensitive info before sharing. Brush or rectangle mode.

B

Crop

Aspect ratio presets (16:9, 4:3, 1:1), orientation toggle, center crop.

C
AI Workflow

Show, don't tell. Review, don't imagine.

Bidirectional visual communication between you and your AI. Integrates with Claude Code, Codex, and Cursor.

You to AI

Show what you see
1

AI asks for a visual -- runs snip-wait to open a URL

2

You capture and annotate -- select region, add arrows/text

3

Copy PNG -- image saved locally, AI reads it instantly

AI to You

Review what it built
1

AI makes changes -- writes code, modifies a page

2

AI requests review -- opens the page via snip-wait

3

You annotate feedback -- circle issues, add notes, send back

Claude Code

Terminal-based AI coding

Codex CLI

OpenAI's coding agent

Cursor

AI-powered IDE

View Setup Guide

Give your AI eyes

Open source. Works locally. No cloud, no accounts, no tracking.

Add to Chrome Setup Guide