Arc Security

This is a security scanner for AI agent workflows that checks your MCP skills and tools against 25 attack categories, from prompt injection to data exfiltration. It runs static analysis on your server implementations and provides runtime monitoring to catch vulnerabilities during execution. The 1,316+ findings reference suggests it's already surfaced issues across a broad corpus of agent code. Reach for this when you're building production agents that handle sensitive data or need to pass security review. It's essentially a linter and runtime guard specifically trained on AI agent attack patterns rather than general application security.