AI RESEARCH
Attested Tool-Server Admission: A Security Extension to the Model Context Protocol
arXiv CS.AI
•
ArXi:2605.24248v1 Announce Type: cross The Model Context Protocol (MCP) standardizes how a large-language-model (LLM) agent and an external tool server exchange messages, but not trust: a host reads a server's self-declared tool list and dispatches calls, with no notion of which servers it may use, at what sensitivity, or which of a server's tools are in bounds.