Glossary
MCP
Model Context Protocol — An open standard that lets AI applications, including LLM-based agents, connect to external tools, data sources, and services in a consistent way. Because it grants models the ability to invoke real external actions, poorly secured MCP integrations expand an AI agent's attack surface and blast radius if compromised via prompt injection.
Recent coverage mentioning MCP
CVE-2026-59822: BerriAI LiteLLM Authentication Bypass
BerriAI LiteLLM is vulnerable to an improper authentication flaw (CVE-2026-59822) actively exploited to bypass authentication.
Securing Model Context Protocol (MCP) Traffic with Cloudflare
Learn how Cloudflare One identifies inspected Model Context Protocol traffic and controls AI agent tool calls to secure enterprise environments.
Unit 42: AI Enhances Attack Efficiency, Not Novel TTPs
Unit 42's report reveals AI accelerates attacker operations, shortening attack lifecycles without fundamentally changing TTPs.
Google ADK for Python RCE: Agent-to-Agent Attacks Expose Secrets
Pillar Security uncovered agent-to-agent RCE flaws in Google's ADK for Python, allowing secret exposure and PR tampering, risking supply chain integrity.
Advertisement