AgentShield Security Scanner

Example Use Cases

• 1

  Verify the safety of MCP servers, skills, or plugins before installation.

• 2

  Detect potential security risks such as backdoors, data theft, and prompt injection.

• 3

  A quick, multi-engine security report to ensure code integrity and trustworthiness.

Description

AgentShield provides a comprehensive security check for AI-related MCP servers, skills, or plugins by running 13 independent scanning engines. It detects risks such as skill hijacking, backdoors, remote control, data theft, prompt injection, tool poisoning, obfuscated code, vulnerabilities, secret leaks, and excessive permissions. The tool aggregates results into a unified report, enabling users to quickly assess the safety of code before installation. It is easy to use with a one-time setup and supports multiple output formats including JSON, HTML, and SARIF.