Read My Screen

Read My Screen is a Raycast extension that turns screenshots, clipboard images, image files, and the active browser tab into AI-powered explanations — using OpenAI (GPT-4o and more), Anthropic Claude, or Google Gemini. You add your own API keys; requests go directly to the provider you choose (no custom backend).

Screenshots

Extension commands in Raycast.

Configure capture, model, and instructions before running analysis.

Analysis in progress with your chosen provider.

Structured reply, copy actions, and follow-up in the same thread.

Reopen past runs from Session History once you have saved conversations.

What you can do

• Screenshot AI — Capture a region, window, or full screen and ask questions (OCR, errors, UI review, summaries).
• Browser tab — Read the current page as text (Chrome, Safari, Arc, Dia, Brave, Edge, Opera, Vivaldi, and more).
• Clipboard & files — Quick commands for clipboard images; pick a local PNG/JPEG/WebP/GIF for analysis.
• Chat — Follow-up questions, regenerate answers, optional model override per session.
• History — Saved sessions with thumbnails; reopen and continue.

Requirements

• macOS with Raycast
• At least one API key:
  • OpenAI
  • Anthropic
  • Google AI Studio (Gemini)

Development

npm install
npm run dev

Build and lint:

npm run build
npm run lint

Versioning (Changesets): The semver lives in package.json (version). After a change worth releasing, run npm run changeset to record it, then npm run version-packages to bump the version and update CHANGELOG.md. Publish to the Raycast Store with npm run publish when ready.

Icon: The menu bar icon is generated from assets/extension-icon.svg. After editing the SVG, run npm run generate-icon (writes extension-icon.png for Raycast).

Configuration

Open Read My Screen in Raycast → Extensions → Preferences to set API keys, default model (must match the provider whose key you use), and default instructions for new runs.

Security & privacy

• API keys stay in Raycast preferences on your Mac. They are not sent to the extension author and are not hardcoded in this repository.
• Network traffic goes directly from your machine to OpenAI, Anthropic, or Google (official APIs), depending on the model you choose. There is no separate backend operated by this project.
• Content you analyze (screenshots, clipboard images, fetched page text, chat messages) is sent to that provider under their terms and your account. Use only data you are allowed to share.
• Session history and custom presets are stored locally via Raycast (and image files under the extension support path).

For reporting security issues in this codebase, see SECURITY.md.

License

MIT — see LICENSE.