explore-vs-exploit-state-machine-skill

What problem does it solve?

Governs when an AI agent should continue gathering information versus acting on the best current option, preventing both premature commitment and endless search.

Core Features & Use Cases

• State-machine protocol with Intake, Uncertainty Assessment, Targeted Exploration, Decision Checkpoint, Exploitation, and Stop states to gate actions.
• Mandatory diagnostic artifact creation (search-budget.md) and explicit stopping criteria to ensure auditable decisions.
• Tool-gating guidance that allows exploration tools, but restricts them during exploitation to prevent analytics drift or unsafe actions.

Quick Start

Instruct the agent to begin with intake, assess uncertainty, perform targeted exploration within a budget, and decide when to exploit.