Agent Audit
Scans MCP server configurations and AI agent tooling for common security vulnerabilities like prompt injection, command injection, and hardcoded secrets.
Agent Audit is a critical security scanner designed specifically for MCP (Model Context Protocol) servers and AI agent tooling. It helps developers and organizations proactively identify and mitigate severe risks such as prompt injection, command injection, hardcoded secrets, and excessive permissions within their MCP server configurations. By understanding the unique attack vectors associated with AI agents and their tools, Agent Audit helps prevent exploits that traditional security tools often miss, ensuring the integrity and safety of your AI deployments.
Key Features
01Detects Command Injection (OWASP A03) and Credential Exposure (OWASP A07)
02Can run as an MCP server itself, enabling direct auditing via AI agents
03Integrates with CI/CD pipelines via GitHub Actions and programmatic API
04Identifies Excessive Permissions (OWASP A05) and Auth Bypass vulnerabilities
05Supports scanning Claude Desktop, Cursor, and custom JSON MCP configs
061 GitHub stars
07Comprehensive scanning for Prompt Injection (OWASP A01)
Use Cases
01Perform pre-deployment security audits of MCP server configurations.
02Integrate into CI/CD pipelines to continuously monitor for new vulnerabilities.
03Proactively scan existing AI agent tooling and MCP setups for security risks.