Kaur-bit/PhishGuard

# PhishGuard A phishing email analyzer that combines AI-powered detection with rule-based pattern matching to instantly assess whether an email is legitimate or a phishing attempt. Live Demo: https://claude.ai/public/artifacts/bb4a51f6-e7ff-436a-ad56-37c06a5550bf What It Does Users paste a suspicious email (or fill in the fields manually) and PhishGuard returns: A risk score from 0–100 A verdict: Safe, Suspicious, or Phishing Detected A breakdown of every red flag found, labeled by severity (High / Medium / Low) A plain-English AI summary of the findings How It Works PhishGuard runs two detection engines simultaneously and combines the results: Rule-Based Engine (40% weight) Scans for known phishing patterns including: Urgency and pressure language Credential harvesting attempts Threatening language Sender domain mismatches Shortened or insecure URLs Brand name spoofing (character substitution) Generic greetings AI Engine via Claude API (60% weight) Sends the email to Claude for contextual analysis catches nuanced threats that simple pattern matching would miss and returns structured findings with severity ratings. Security Features Input sanitization : strips all HTML tags and JavaScript before processing Prompt injection protection : user-supplied email content is wrapped in hard delimiters and the AI is explicitly instructed to treat all content as untrusted data, not instructions Character limits : enforced on all inputs to prevent abuse Tech Stack React Claude API (claude-sonnet-4-20250514) Vanilla CSS with CSS custom propertie _AI-powered phishing email analyzer built with React and Claude API, combines rule-based pattern detection with live AI analysis to score and break down suspicious email_

标签：自定义脚本