validation-criteria

You establish binary evaluation criteria by identifying explicit pass/fail conditions for AI outputs. This framework enforces a binary-testability schema to ensure consistent QA and capture validated examples in YAML format.

The best way to structure QA criteria for AI governance is to apply a guided collection flow that elicits specific, testable conditions. This aligns outputs with a YAML schema to enable consistent feedback cycles and machine-checkable validation.

Validated AI evaluation examples are saved as YAML files within the .claude/validations directory. This structured storage ensures criteria are formatted according to the defined schema for future QA and governance workflows.

You enforce a binary-testability schema by using push-back prompts during a guided collection flow. This process requires articulating observable criteria that provide a definitive pass or fail assessment for AI outputs.

Yes, binary validation criteria can be applied across different AI tasks and skills. This meta-skill framework ensures consistent evaluation, example capture, and governance of feedback cycles regardless of the specific domain.