decision-policy

What problem does it solve?

This Skill establishes a clear, auditable policy for agent decision-making, ensuring that critical or risky actions are either automatically approved based on predefined rules or explicitly escalated for user confirmation.

Core Features & Use Cases

• Decision Classification: Categorizes pending decisions into types like security, cost, or destructive actions.
• Risk Assessment: Evaluates decisions against dimensions such as blast radius, reversibility, and data sensitivity using an escalation matrix.
• Escalation Enforcement: Automatically approves low-risk, reversible actions while mandating user approval for high-risk, irreversible, or sensitive operations.
• Decision Logging: Records the decision, rationale, risk scores, and outcome for auditability.
• Use Case: When an agent needs to modify security configurations or perform a potentially costly operation, this Skill will first assess the risk and then either proceed autonomously or request your explicit approval.

Quick Start

Use the decision-policy skill to classify a pending change to the application's authentication flow and determine if user approval is required.