confirmation-final-qa

name: confirmation-final-qa description: Create confirmation summary and final quality checks. ---# Confirmation and Final QA Skill

Purpose

Create the user-facing confirmation summary and block unsafe or incomplete submission.

When to use

Use after scoring is complete and before any Dataverse write.

Inputs

- Working object. - Assessment output. - User intent: Submit, Edit, Cancel, or unknown.

Confirmation summary must include

- Customer Name - Idea Title - Use Case Category - Use Case Subcategory - Summary - Problem / Need - Proposed Solution - Expected Benefits - Target Audience - Personas Impacted - Potential Risks - Sample Prompts, Agent Questions, and Example Answers - Impact Assessment table: Factor | Score | Rationale - Complexity Assessment table: Factor | Score | Rationale - Reach Score - Agentic Fit - Impact Average - Complexity Average - Weak spots / validation notes - Missing or NOT ENOUGH INFO areas - Recommended questions to complete the assessment

Rules

- Ask: Ready to record the findings? - Suggested replies: 1. Submit 2. Edit 3. Cancel - Tell the user they can reply with the number or the option text. - If Edit, ask only which fields to change and re-confirm after edits. - If Cancel, end politely without saving. - If Submit, validate before save. - Do not expose table names, columns, schemas, GUIDs, or duplicate-check logic. - Do not reveal CAD Intake Quality. - Do not expose internal rubric codes or factor IDs such as I1, I2, I5, C1, or similar. - Use plain factor names only in confirmation tables. - Use "Use case area" and "Specific use case type" as the user-facing labels for category and subcategory. - If this is a provisional assessment, clearly label it as provisional and show the unresolved gaps before asking for Submit. - Use compact tables for confirmation summaries where helpful. - Required field confirmation can use columns: Field | Value | Confidence / note. - Include sample prompts, agent questions, and example answers in the confirmation summary. If missing in provisional mode, show it as a gap and recommended follow-up. - Impact and Complexity assessments must use columns: Factor | Score | Rationale. - Provisional gaps can use columns: Area | Status | What is missing | Recommended follow-up.

Blocking validation

Block submission if:

- Any required field is missing. - Category/subcategory is NOT ENOUGH INFO. - Any factor is NOT ENOUGH INFO in Guided Completion or Full Assessment mode. - Reach Score is missing or outside the allowed range. - Averages are missing or inconsistent. - User did not explicitly choose Submit. - Dataverse save requirements cannot represent NOT ENOUGH INFO values in provisional mode.

Allow submission in Provisional Assessment mode only if:

- The user explicitly chose Run provisional assessment. - All missing values are clearly marked NOT ENOUGH INFO. - The report will include the missing evidence summary and recommended questions. - Dataverse can store unresolved fields without invalid placeholder scores.

Output contract

{
  "confirmationSummary": "",
  "readyToSubmit": false,
  "blockingIssues": [],
  "editPrompt": "",
  "cancelMessage": ""
}