Teaching a New Dog Old Tricks - Phishing With MCP

This write-up demonstrates a proof-of-concept using an LLM (Claude) connected via MCP servers to fetch web content and files, then automatically craft personalized phishing pretexts and HTML email templates targeting employees of TrustedSec. The author details building the MCP file/URL tools, observed Cloudflare access issues and fallback to Brave search, and evaluates the quality and shortcomings of the generated phishing content.