Guardian FAQs

Question 1

What types of security threats does Guardian detect?

Accepted Answer

Guardian detects a wide range of threats including prompt injection, command injection, PII exposure, excessive permissions, dangerous data exfiltration chains (the 'Lethal Trifecta'), missing input validation, and tool poisoning.

Question 2

What is the 'Lethal Trifecta' and how does Guardian address it?

Accepted Answer

The Lethal Trifecta describes a common vulnerability where seemingly harmless individual tools (e.g., file access, internet processing, email sending) combine to form a complete data exfiltration chain. Guardian identifies these dangerous patterns and shuts them down by applying targeted blocking policies.

Question 3

What is Guardian and its main purpose?

Accepted Answer

Guardian is an MCP server designed for the Archestra platform. It automatically scans, tests, and locks down other MCP servers to prevent critical vulnerabilities, data exfiltration, and AI agent exploits.

Question 4

How does Guardian provide automated security for MCP servers?

Accepted Answer

It performs comprehensive vulnerability scanning (including 50+ patterns and LLM deep scans), automated security testing for injection and edge-case exploits, and real-time monitoring. Crucially, it generates and applies Archestra security policies to enforce protection.

Question 5

How does Guardian prevent vulnerabilities after detection?

Accepted Answer

Unlike tools that only report, Guardian implements 'closed-loop security.' It automatically generates and applies Archestra Tool Invocation and Trusted Data Policies directly to the platform, blocking dangerous actions or sanitizing outputs before AI agents can exploit them.

Guardian

Guardian

主な機能

ユースケース

主な機能

ユースケース