A2M: Trace-Optimized Agent Hijacking in the MCP Ecosystem

T1Covered by 1 source · first reported by arXiv at 24 Sept, 04:00 UTC

arXiv:2609.26761v1 Announce Type: cross Abstract: Agents using the Model Context Protocol (MCP) rely on semantic matching to select tools from third-party servers, exposing a semantic supply-chain risk through attacker-controlled metadata and outputs. We introduce A2M (Attraction-to-Manipulation)…

science

Sources

• T1A2M: Trace-Optimized Agent Hijacking in the MCP EcosystemarXiv24 Sept, 04:00