analiza aiqbench

This high-severity vulnerability in the Model Context Protocol (MCP) Python SDK exposes AI agent accounts to hijacking by allowing malicious servers to manipulate OAuth flows and steal access tokens from legitimate identity providers. The flaw stems from the SDK's failure to validate OAuth issuer configurations when falling back to legacy paths, enabling attackers to intercept authorization codes and PKCE verifiers while presenting a deceptive login interface to users. For developers integrating MCP with external tools, this incident underscores the critical need for rigorous validation of authentication metadata in untrusted environments to prevent credential theft within AI agent ecosystems.