pragmatic-programmer-state-machine-skill

To manage blast radius, you enforce public interface discovery, unknowns declaration, and bounded execution before an AI agent makes changes. This ensures modifications are reversible and risks are minimized during execution.

A bounded state-machine workflow guides AI agents through intake, consumer discovery, option selection, execution, and validation. It enforces clean stopping conditions and risk-management constraints to ensure reliable, reversible modifications.

You ensure AI agent changes are reversible by applying blast-radius accounting and bounded execution constraints. This pragmatic approach requires declaring unknowns and defining clean stop conditions before execution begins.

No dependencies or external components are required. You simply describe the bounded change you want implemented, and the skill guides the discovery, option selection, and execution workflow natively.

You reduce manual toil by scripting repeatable, bounded improvements through the state-machine workflow. This automates intake, execution, and validation to deliver root-cause fixes with quick turnaround.

You should avoid bounded state-machines when execution speed is prioritized over reversibility and blast-radius accounting. The workflow enforces discovery and validation steps that add overhead unsuitable for unstructured tasks.