Speak a command, and Claude AI sees your screen, understands the full page via Firecrawl, and executes actions autonomously. Search the web, fill forms, navigate sites — all by voice. Powered by Claude Haiku 4.5, ElevenLabs voice, and Firecrawl web scraping.
No tab switching, no copy-pasting, no context loss. Just hold, ask, and understand.
Press and hold the shortcut key anywhere on any webpage. A waveform appears — you're live.
"Add Quest protein bars to my Amazon cart." "Search YouTube for coding tutorials." "Scroll down and find the cheapest option."
Claude AI sees your screen, Firecrawl reads the full page, and the agent clicks, types, navigates — autonomously completing your task step by step.
Watch how to install, configure, and start talking to your screen.
Powerful when you need it, invisible when you don't.
Multimodal AI sees your screenshot + DOM structure. Understands page context, identifies elements, and plans actions with chain-of-thought reasoning.
Hold backtick, speak naturally. Groq Whisper converts speech to text with ElevenLabs voice readback, then Claude decides and acts on your behalf.
After each action, Claude re-captures the screen, Firecrawl re-scrapes the full page content, and decides the next step. Loops up to 25 iterations until your task is complete.
Failed selectors? Claude retries with fresh DOM. Navigation mid-task? Waits for reload and continues. Element highlighting shows exactly what's being clicked.
Ask "What is an API?" and watch the answer adapt to who's asking.
Install ScreenSense Voice for free. Speak a command on any website — Claude + ElevenLabs + Firecrawl handles the rest.
Clone from GitHubClone the repo, run npm install && npm run build, then load the dist/ folder into chrome://extensions with Developer Mode enabled. Follow the README setup guide for detailed steps.