Skip to content
Michael Bargury· @mbrg0 · X·Indexed by this site · Originally published

Zenity Labs discloses an attack against AWS Bedrock AgentCore: a single prompt steals execution role credentials

AI summary

Zenity Labs says a single prompt can compromise an exposed AWS Bedrock AgentCore Agent because IMDS is not shielded inside the microVM; one HTTP request stole execution role credentials. Author Michael Bargury retweeted and called it a particularly tricky case. The research points the attack surface to metadata service access control in cloud agent runtime environments; the specific affected configuration scope and fix status are not stated in the material.

Full text · Original

this was a particularly nasty one. if you're building agent in the cloud.. sorry and enjoy! :)

@_d1voy @lana__salameh @tamirishaysh @p1njc70r

Quoted postzenitylabs@zenitysec_labs
New research: one prompt to one exposed AWS Bedrock AgentCore agent collapsed the entire cloud security model. IMDS wasn't blocked inside the microVM. One HTTP request stole the execution role credentials. #ZenityLabs #AgentSecurity #AWS
View the quoted post on X

Source: Michael Bargury · x.com