threat-intel
Malicious MCP Servers Can Split Instructions to Make AI Coding Agents Exfiltrate Secrets
High
Summary
A malicious tool server leveraging the Model Context Protocol (MCP) can silently exfiltrate sensitive data – including SSH keys, source code, and customer data – from AI coding assistants. The attack works by splitting requests into fragments and letting the agent stitch them together, even if individual fragments appear harmless. The technique, dubbed GhostSplice, highlights a vulnerability in how AI assistants interact with external tools and emphasizes the importance of client-side security measures.
Summary written automatically in our own words from the original article, which belongs to its publisher and remains the reference. It may contain errors. Sources & data