MCP Attacks

Test Name	API ID	Description	Detectable Vulnerabilities
AESI: ANSI Escape Sequence Injection	`aesi`	Checks whether MCP tools, resources, or prompt templates can fetch external content and return ANSI-concealed instructions that are hidden from terminal users but still visible to downstream models	AESI: ANSI Escape Sequence Injection
Stored AESI: ANSI Escape Sequence Injection	`stored_aesi`	Checks whether MCP integrations can persist externally hosted ANSI-concealed instructions and expose them later to downstream models	Stored AESI: ANSI Escape Sequence Injection