Model Context Protocol Threat Auditor
Audit MCP Servers for Security Vulnerabilities
Detect command injection, unauthenticated SSE transports, hardcoded secrets, and untrusted input vulnerabilities before exposing tools to production runtimes.
Try sample:
Command Injection Guard
Flags dangerous shell primitives (`subprocess`, `os.system`, `eval`) where untrusted agent tool calls could execute host commands.
Transport Authentication
Verifies that SSE and HTTP JSON-RPC transports enforce token or API-key authorization headers to prevent open network access.
Prompt Injection Mitigation
Audits parameter schemas and descriptions to prevent jailbreak attacks from forcing arbitrary tool invocation or exfiltration.