policy

What problem does it solve?

This Skill provides a structured, deterministic way to guide an AI agent's decision-making process, ensuring safety, consistency, and adherence to predefined operational logic, moving beyond purely LLM-driven choices.

Core Features & Use Cases

• Rule-Based Decision Making: Define codified rules (preconditions and effects) that the agent evaluates before executing actions.
• Safety and Guardrails: Implement high-priority rules to block dangerous commands or prevent undesirable outcomes.
• Behavioral Control: Influence action selection through boosting, filtering, or rewriting candidate actions.
• Impasse Handling: Define escalation paths when the agent becomes stuck or encounters repeated failures.
• Use Case: An agent needs to access a file system. You can define a policy rule that blocks any rm -rf / command, ensuring system safety, while another rule might boost a specific search skill when the task involves research papers.

Quick Start

Use the policy skill to evaluate a set of candidate actions against the current state and apply predefined rules.