Generate Customer Qa.prompt

# 60 - Generate Customer / Domain Expert Q&A

## Purpose

Generate a concise customer/domain expert Q&A list for unresolved items that cannot be safely finalized from the current FD evidence.

Use this step only when the latest FD review or patched FD review indicates that clarification is needed before the FD can be considered safe for customer/domain expert review, DD, coding design, or testing.

This task must not generate, revise, or patch the FD.

## Operating rules

- Use only the source artifacts listed in this prompt.
- Do not invent requirements, business rules, messages, screen behavior, command behavior, API behavior, or design decisions.
- Ask only questions that require customer/domain expert clarification.
- Do not ask questions for issues that can be fixed internally from existing source evidence.
- Do not ask questions caused only by AI writing style, formatting, or traceability-table quality.
- Do not expose internal artifact names or internal IDs in the customer-facing Q&A.
- Rewrite internal findings into natural customer-facing questions.
- Preserve source-specific terms, product terms, command names, option names, field names, acronyms, and code-like identifiers exactly unless the translation policy or glossary defines a safe rendering.
- Follow `output/11_translation_policy.md` and `output/21_glossary.md` for terminology control.
- Respect strikethrough/deprecated evidence rules: strikethrough-only content must not become an active behavior question unless it conflicts with active evidence or the source explicitly says otherwise.
- Write the Q&A in English.

### Important

- This is a clarification artifact, not an FD correction task.
- This task should filter questions aggressively.
- The output should be short enough to send to a customer/domain expert.
- If an issue can be corrected internally using `output/30_requirement_inventory.md`, `output/31_business_rule_catalog.md`, `output/32_open_questions.md`, `output/21_glossary.md`, `output/20_image_analysis.md`, or `output/12_normalized_evidence.md`, do not ask the customer.

## Tasks

### Precondition

Use the latest available FD review gate in this order:

1. If `output/55_FD_PATCHED_REVIEW_REPORT.md` exists, use it as the latest review source.
2. Otherwise, if `output/52_FD_REVISED_REVIEW_REPORT.md` exists, use it as the latest review source.
3. Otherwise, use `output/42_FD_REVIEW_REPORT.md`.

Generate Q&A when the latest review result is:

- `No-Go`, or
- `Go with minor corrections`, or
- `Go` but the report explicitly recommends customer/domain expert Q&A.

If the latest review result is `Go` and no customer/domain expert Q&A is recommended, create the output file with a short summary stating that no customer/domain expert questions are required at this time.

### Instructions

1. Review unresolved issues from the latest review report.
2. Cross-check against previous review reports when relevant so repeated unresolved issues are not lost.
3. Cross-check against source-of-truth artifacts to remove questions that can be answered internally.
4. Convert only real source ambiguity or missing source detail into customer/domain expert questions.
5. Group questions by priority and topic.
6. For each question, include a customer-facing source reference.
7. Provide a safe suggested default handling only when a safe default exists.
8. If no safe default exists, use `TBD`.
9. Do not expose internal IDs such as `REQ-xxx`, `BR-xxx`, `TERM-xxx`, `OQ-xxx`, `FIG-xxx`, `TRACE-xxx`, `RTRACE-xxx`, or `PTRACE-xxx`.
10. Do not expose internal artifact names such as requirement inventory, business rule catalog, glossary, image analysis, translation review, traceability, intermediate artifact, prompt, AI, or pipeline.

## Inputs

### Primary inputs

Use the latest available FD and review set:

- `output/55_FD_PATCHED_REVIEW_REPORT.md`, if available
- `output/54_FD_PATCHED_INTERNAL_TRACEABILITY.md`, if available
- `output/53_FD_DRAFT_REVISED_PATCHED.md`, if available
- `output/53_FD_PATCH_LOG.md`, if available

If patched FD artifacts are not available, use the revised FD set:

- `output/52_FD_REVISED_REVIEW_REPORT.md`, if available
- `output/51_FD_REVISED_INTERNAL_TRACEABILITY.md`, if available
- `output/50_FD_DRAFT_REVISED.md`, if available
- `output/50_FD_REVISION_LOG.md`, if available

If revised FD artifacts are not available, use the first FD review set:

- `output/42_FD_REVIEW_REPORT.md`
- `output/41_FD_INTERNAL_TRACEABILITY.md`
- `output/40_FD_DRAFT.md`

### Source-of-truth inputs

Use these to decide whether a question is genuinely needed or can be resolved internally:

- `output/30_requirement_inventory.md`
- `output/31_business_rule_catalog.md`
- `output/32_open_questions.md`
- `output/21_glossary.md`
- `output/20_image_analysis.md`
- `output/12_normalized_evidence.md`

### Supporting inputs, open only when needed

- `output/11_translation_policy.md`
- `output/10_document_inventory.md`
- `working/extracted/document_text.md`
- `working/extracted/tables.md`
- `working/extracted/image_inventory_raw.md`

## Outputs

### Output file to create or update

Create or overwrite:

- `output/60_CUSTOMER_QA.md`

## Required output structure

Write `output/60_CUSTOMER_QA.md` using this structure: