Apply fd Review Corrections.prompt
# 50 - Apply FD Review Corrections
## Purpose
Apply first-round review corrections to the customer-facing Functional Design (FD) without introducing unsupported content.
This task patches the FD generated in step 40 using the review result from step 42 and the source-bound evidence artifacts from steps 12, 20, 21, 30, 31, and 32.
This task must produce a complete revised FD document, not a diff.
## Operating rules
- Use only the source artifacts listed in this prompt.
- Do not regenerate the full FD from scratch unless `output/42_FD_REVIEW_REPORT.md` explicitly says the FD structure is unusable.
- Do not invent requirements, business rules, validations, command behavior, UI/API behavior, messages, diagrams, or implementation design.
- Do not introduce implementation architecture unless the source explicitly supports it.
- Preserve source-supported conditions, exceptions, notes, footnotes, negative statements, unsupported behavior, and visual evidence.
- Preserve command names, option names, parameters, field names, product-specific terms, acronyms, and code-like identifiers exactly.
- Keep the revised FD customer-facing.
- Do not expose internal artifact names, internal IDs, review issue IDs, or pipeline terminology in the customer-facing FD body.
- Internal review details may appear only in `output/50_FD_REVISION_LOG.md`, not in the customer-facing FD body.
- Write outputs in English unless a source value must remain unchanged.
### Customer-facing cleanliness rules
The customer-facing FD body must not mention or expose:
- internal artifact file names
- `REQ-xxx`
- `BR-xxx`
- `TERM-xxx`
- `OQ-xxx`
- `FIG-xxx`
- `TRACE-xxx`
- review issue IDs
- requirement inventory
- business rule catalog
- glossary-controlled
- image analysis
- translation review
- internal traceability
- intermediate artifact
- prompt, model, Copilot, or AI pipeline wording
### Strikethrough and deprecated evidence rules
- Strikethrough text represents inactive, deleted, deprecated, superseded, or draft-removed evidence unless the source explicitly states otherwise.
- Do not use strikethrough-only content as active FD behavior.
- Do not use strikethrough-only content as an active business rule, active operation flow, active command syntax, active UI/API behavior, or active field definition.
- If the existing FD used strikethrough-only content as active behavior, remove it, mark it as not confirmed, or move the uncertainty to Open Questions.
- If strikethrough content conflicts with active evidence, preserve the active evidence and add an Open Question only when the conflict affects behavior.
## Tasks
### Precondition
Use the latest FD review gate from `output/42_FD_REVIEW_REPORT.md`.
Continue only when the step 42 decision is one of:
- `Go with minor corrections`
- `Fix required`
- `No-Go` where the review report clearly identifies fixable corrections
Stop if:
- `output/42_FD_REVIEW_REPORT.md` is missing.
- `output/40_FD_DRAFT.md` is missing.
- the review report says the FD cannot be safely corrected from available evidence.
- applying corrections would require unsupported assumptions.
### Instructions
Apply targeted corrections in this priority order:
1. Fix Critical findings from `output/42_FD_REVIEW_REPORT.md`.
2. Fix Major findings from `output/42_FD_REVIEW_REPORT.md`.
3. Remove customer-facing cleanliness issues.
4. Remove or rewrite untraced FD statements.
5. Weaken potential overstatements to match the source evidence.
6. Add missing source-supported requirement/rule coverage.
7. Move uncertain source-supported items to Open Questions.
8. Fix terminology inconsistencies.
9. Fix visual traceability issues.
10. Fix minor wording and formatting issues.
For each correction:
- Use the original FD wording as the starting point.
- Patch only the affected section unless a broader restructure is explicitly required by the review report.
- Keep confirmed source-supported behavior as confirmed behavior.
- Convert unsupported or uncertain behavior into `TBD`, `Requires confirmation`, or an Open Question.
- Remove editorial, generic, marketing-like, or untraced statements that do not help the FD.
- Do not add new behavior just because it seems reasonable.
## Inputs
### Primary inputs
- `output/40_FD_DRAFT.md`
- `output/41_FD_INTERNAL_TRACEABILITY.md`
- `output/42_FD_REVIEW_REPORT.md`
- `output/30_requirement_inventory.md`
- `output/31_business_rule_catalog.md`
- `output/32_open_questions.md`
- `output/21_glossary.md`
### Supporting inputs, open only when needed
- `output/20_image_analysis.md`
- `output/12_normalized_evidence.md`
- `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 files to create or update
- `output/50_FD_DRAFT_REVISED.md`
- `output/50_FD_REVISION_LOG.md`
## Required output structure
### Output 1: `output/50_FD_DRAFT_REVISED.md`
Create a complete revised customer-facing FD document.
The revised FD must:
- keep the same overall customer-facing FD style as `output/40_FD_DRAFT.md` unless the review report requires structural correction.
- be a full document, not a diff.
- not expose internal IDs or internal artifact names.
- include only source-supported behavior.
- mark uncertain behavior as `TBD`, `Requires confirmation`, or Open Question.
- preserve source-supported unsupported cases, negative statements, conditions, exceptions, notes, and footnotes.
- preserve source-supported visual content only when traceable.
Do not include an internal patch summary inside `output/50_FD_DRAFT_REVISED.md`.
### Output 2: `output/50_FD_REVISION_LOG.md`
Create an internal-only revision log using this structure:when to use it
Community prompt sourced from the open-source GitHub repo tuana-vn/brd-to-fd-copilot-agent (MIT). A "Apply fd Review Corrections.prompt" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.
tags
productivitycommunitydeveloper
source
tuana-vn/brd-to-fd-copilot-agent · MIT
more in Productivity
Productivity✓ tested
Summarize a doc into decisions & actions
chief of staff who extracts what to DO, not just what was said
Productivity✓ tested
Draft a reply to a hard email
calm, direct communicator who de-escalates without caving
Productivity✓ tested
Turn a brain-dump into a weekly plan
planning coach who protects your focus, not just your calendar