About
In this module of the Advanced AI Security Part 2 learning path, you will treat MCP tools as an HTTP client and a code interpreter: tool SSRF, code execution, and output sinks on the same wrapper. `fetch_url` has no allowlist, so loopback IMDS is in scope. `run_python` allows `open()` and `print()`. `/render` drops model output into HTML. `lookup_student` concatenates names into SQL. Those are LLM03:2026 (excessive agency) and LLM10:2026 (improper output handling) on the same wrapper. Upon completion of the module, students will be able to: - Call fetch_url against http://127.0.0.1:8080/internal/imds. - Read the shell ticket with run_python. - Show the HTML comment token on /render and a UNION-style SQL read. For inquiries, please write us at https://www.darkrelay.com/cybersecurity-course-inquiry