This project provides the Chrome Extension version of the Deep Think service β an AI-powered tool for critical reading and media analysis. The extension enables users to analyze web content directly within the page they're viewing, offering sentence-level classification, summary, and author intent detection without switching tabs or copying text.
-
One-click analysis of any webpage without copying or pasting
-
In-page overlays that highlight sentences based on clarity:
- π© Clear sentences: Objectively interpretable or fact-based
- πͺ Ambiguous sentences: Subjective, emotional, or open to multiple interpretations
-
summary and author intent at the top of the page
-
Interactive tooltips on sentence click, with:
- Classification explanation
- Alternative interpretations (if ambiguous)
- Reference links to external sources (if ambiguous)
-
Popup shortcut that lets users analyze the current webpage in the full Deep Think web app β no need to copy or paste anything.
- Navigate to any article or webpage.
- Click the Deep Think floating button at the top-right corner of the page.
-
The AI will:
- Display a summary and author intent analysis at the top.
- Highlight all sentences as clear or ambiguous.
- Show explanation tooltips when you click on a sentence.
- Click the Deep Think icon in the Chrome toolbar.
- A popup will appear with an option to open the current page inside the Deep Think web app.
- This launches a full-page version of the analysis with enhanced UI and controls.
To run this extension locally for development:
-
Clone the repo and open
chrome-extensiondirectory. -
Run the local build:
npm install npm run build -
Open chrome://extensions/ in Chrome.
-
Enable Developer Mode (top right).
-
Click βLoad unpackedβ and select the
dist/directory.
Important Configuration:
- In
content.js, update theonclickfunction to point to your server address. - In
extension.js, update theextensionfunction to point to your client address.
This extension is part of the Deep Think project β helping users engage with digital content more critically by leveraging explainable AI and in-context feedback.
MIT License




