agentic-design-patterns-orchestrator-state-machine-skill

Multi-step AI workflow orchestration requires a state-machine protocol to classify tasks, select patterns, gate execution, and verify outputs across phases. This prevents unstructured single-shot responses by enforcing explicit planning, evidence gathering, and safe termination conditions.

A state-machine orchestration protocol structures AI task execution into explicit phases including classification, pattern selection, planning, execution gating, reflection, and verification. It generates diagnostic artifacts like an agentic-run-plan.md to ensure repeatable, gated multi-step workflows.

Managing unknowns in complex AI task execution involves maintaining an unknowns-register and applying phase-based tool permissions. This risk-management approach captures gaps during planning and gates tool access to ensure diagnostic evidence gathering before execution proceeds.

A stateful workflow is needed when AI tasks require multi-pattern coordination, execution gating, diagnostic evidence gathering, and explicit stop conditions. Complex workflows that demand structured phase-based verification rather than immediate generation benefit from this orchestration approach.

Execution gating and tool permissions are configured through the orchestrator state machine by defining explicit phases in the agentic-run-plan. Each phase restricts tool access, requiring diagnostic evidence and reflection before advancing to execution or verification states.

Yes, a state machine handles complex AI workflows by defining explicit stop conditions within the orchestration protocol. It gates task execution across phases, ensuring safe termination only after verification and reflection criteria are fully satisfied.