guardrail

What problem does it solve?

This Skill acts as a critical policy enforcement layer, preventing AI agents from executing unauthorized or unsafe actions by validating intent, roles, and action categories.

Core Features & Use Cases

• Intent Validation: Ensures agents declare read_only or mutating intent correctly.
• RBAC Authorization: Verifies user roles and approval permissions for mutating actions.
• Mutating Category Check: Classifies mutating actions into predefined security categories.
• Blast Radius Gate: Enforces elevated approval for high-impact actions.
• Use Case: When an agent proposes to modify a firewall rule, Guardrail checks if the user has the admin role, if an approver with the correct role exists, and if the action is correctly categorized as policy_change before allowing it to proceed.

Quick Start

Use the guardrail skill to validate a proposed mutating action for a firewall rule change.