Marlint - AI Phishing & Malicious Assessor
Unknown author · Privacy & Security
Media
2 items
Description
Assess visited websites for phishing and malicious indicators using OpenRouter or local LM-Studio. Marlint: Real-Time AI Phishing & Malicious Site Detection Protect your browsing experience with Marlint, a lightweight, real-time security extension that leverages the power of Large Language Models (LLMs) to detect phishing, fraudulent clones, scams, and deceptive websites before they can compromise your data. By analyzing domain structures, subtle typosquatting patterns, and misleading URLs in real-time, Marlint blocks malicious sites and helps you navigate the web with confidence. Key Features 1. AI-Powered Threat Assessment: Detects sophisticated phishing, fake service clones (like spoofed banking or brand login portals), and deceptive content using advanced LLM reasoning. 2. Dual-Engine Flexibility: - OpenRouter (Cloud): Connect to any cloud-based instruct model using your own API key. - LM-Studio (Local/Offline): Connect to a model running on your local machine for 100% private, offline, zero-data-sharing classification. 3. Privacy-Enhancing Tranco Safe-List: Marlint automatically downloads, unzips, and indexes the Tranco Top 1 Million popular websites locally inside your browser. Known-good, globally trusted domains are skipped instantly without any external requests. 4. Local Whitelisting: Put absolute trust in specific websites by adding them to your custom whitelist with a single click. 5. Performance-Focused Caching: Threat assessments are cached locally for 24 hours to maximize loading speeds and minimize API requests. 6. Intuitive Warning Intercepts: Blocks malicious destinations with a clear warning page containing detailed classifications, safety scores (0-100), and AI-generated reasoning. How It Works 1. Smart Filtering: Upon navigating to a website, Marlint immediately checks your local whitelist and the local Tranco Top 1M safe-list. If there is a match, browsing continues instantly. 2. AI Classification: If the domain is unknown, Marlint queries your selected LLM provider (locally or via API) for a safety score. 3. Active Protection: If deemed malicious, Marlint intercepts the tab and displays a blocked warning page detailing the exact threat detected. Note: Marlint is built with extreme privacy in mind. It only analyzes hostnames and URLs. Your page content, body text, form inputs, cookies, and network headers are never accessed or transmitted. Setup & Configuration Configure Marlint in seconds through the extension popup: 1. Choose your provider: Select either OpenRouter or LM-Studio. 2. Enter your API credentials: Input your OpenRouter API key, or use Marlint completely key-free with a locally hosted model on localhost. 3. Toggle Protection: Customize whether the extension auto-checks every website, or is set to manual "on-demand" scan.