Agent Security Scanner

You'd reach for this when you need to security test AI agents that use tool calling. It scans for common vulnerabilities like prompt injection attacks, prompt leaks where system instructions get exposed, and tool hijacking where malicious inputs trick the agent into calling unintended functions. Think of it as a vulnerability scanner specifically built for the unique attack surface of LLM agents with function calling capabilities. The implementation details are light from the source, but the use case is clear: run it against your agent before deployment to catch the kinds of exploits that traditional security testing won't find.