Builds 1.0.137 through 1.18.32 can run a command from a malicious MCP server at sign-in, and no CVE was filed…
An authenticated flaw that passes a user-supplied command straight to bash is being hammered by attackers who only need one…
A malicious MCP server can redirect a client's OAuth handshake and pocket the secrets.
A prompt injection flaw in Amazon's Kiro IDE can push sensitive local data to attackers when a poisoned project is…
Tenet Security researchers discovered a technique called Agentjacking that uses crafted Sentry error reports to trick AI coding assistants into…
An open-source MCP server transforms Claude AI into a security analyst with 27 tools spanning 21 APIs for CVE lookup,…