pdd-to-test-prompts-qa

Structural validation of test prompt markdown checks header alignment, prompt count bounds, required field presence, and full adversarial category coverage before quality evaluation. This skill runs 8 static checks and returns a binary pass/fail verdict with YAML QA results.

Adversarial prompt coverage requires full coverage of 7 adversarial categories and a minimum 20% adversarial share of total prompts. Test prompt suites must also include mandatory training-gap, product-feedback, and escalation prompts to pass structural validation.

Failing structural checks can be fixed using auto-fix hints that provide specific, actionable suggestions for each failed validation. Review the structured YAML QA results to remedy header mismatches, missing required fields, insufficient adversarial counts, or missing mandatory prompt types.

A binary pass/fail verdict gates quality grading to ensure only structurally valid prompt suites move to resource-intensive quality evaluation. This prevents wasted resources on malformed markdown artifacts lacking required fields, sufficient adversarial coverage, or mandatory prompt types.

Structural validation is designed specifically for ACE Connect lifecycle opportunities and supervisor chatbots. The 8 static checks verify requirements tailored to this workflow, including 7 specific adversarial categories and mandatory training-gap, product-feedback, and escalation prompts.

Automated structural validation only checks formatting and coverage requirements like field presence and adversarial share, not semantic quality or prompt effectiveness. It gates the companion quality grading skill but does not evaluate chatbot response accuracy or conversational relevance.