ThreatsDay Bulletin: Worm Code Leaked, AI Agent Phished, Claude Code Patch + 28 New Stories

This bulletin summarizes a high-risk week in threat intelligence: source code for a supply-chain worm was leaked and rapidly forked, a new class of 'AI agent phishing' was demonstrated enabling automated credential exfiltration, Anthropic issued a Claude Code security patch, and a $5,000/month browser-cloning RAT actively in use can fully impersonate user sessions; the report urges treating AI agents as privileged, monitoring public repos for leaked tooling, enforcing session security, and improving supply-chain hygiene.