ferrox-ai-integration-phase

An AI design contract specifies the intended architecture, responsibilities, and constraints for a phase involving AI work. It reduces ambiguity and unverified agent behavior by documenting design and evaluation decisions before implementation starts.

Defining an evaluation strategy before building an AI system reduces ambiguity and unverified agent behavior. This Skill orchestrates evaluation strategy design alongside framework selection to ensure model choice and domain assumptions are specified upfront.

The best way to document framework selection and domain research is through a fixed workflow that produces an AI-SPEC.md. This Skill coordinates research and planning steps to pin down model choice, domain assumptions, and evaluation criteria.

Yes, you can use this workflow planning approach for any product phase requiring an AI feature. It applies to phases that need design and evaluation decisions documented first, coordinating research to produce a pinned-down specification.

An AI design contract should include the intended architecture, responsibilities, constraints, model choice, domain assumptions, and evaluation criteria. This Skill ensures all these elements are specified in an AI-SPEC.md before development begins.