Add AI Agent Risk Self-Check

[MAUROCERON]

Adds AI Agent Risk Self-Check to Guardrails & Compliance.

Why it fits:

• Directly focused on AI agent workflow security.
• Free, no-login, client-side self-check.
• Open-source repo with MIT license and recent maintenance.
• Covers prompt injection boundaries, tool permissions, browser/file access, approval gates, logging, and credential handling.

The public page includes an optional paid mini-audit CTA, but the submitted resource itself is free and usable without payment or account creation.

[secxena]

Thanks for contributing. This PR duplicates #51 from the same repo — please keep a single PR per tool. Beyond that, the project was created only days before submission and has no adoption yet; to keep the list high-signal we hold entries to a bar of demonstrated usage and sustained maintenance. Happy to revisit as it matures.