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.
