Hey HN we’ve been collecting lethal trifecta based attack scenarios on official MCPs and implementing guardrails against them for a while now. It's incredible to see how many of the official MCPs are susceptible to these attacks. With ChatGPT’s integrated MCP support lethal trifecta attacks have become much more relevant.
Noobs using MCP as ai tools get more powerful is going to get really interesting. Incredible that we're already seeing malware that prompts Claude to find things on a host.
2 comments