doc-brd（文档文档）

doc-brd

Purpose

Create Business Requirements Documents (BRD) - Layer 1 artifact in the SDD workflow that defines high-level business needs, strategic objectives, and success criteria.

Layer: 1 (Entry point - no upstream dependencies)

Downstream Artifacts: PRD (Layer 2), EARS (Layer 3), BDD (Layer 4), ADR (Layer 5)

MVP → PROD → NEW MVP Lifecycle

Key Principle: Each BRD represents ONE iteration cycle.

BRD-01 (MVP) → Production v1 → Feedback → BRD-02 (NEW MVP) → Production v2 → ...

Rules:

• New features = New BRD (don't expand existing BRDs indefinitely)
• Each BRD should have 5-15 focused requirements
• Link cycles via Cross-BRD Dependencies (@depends: BRD-01)

Prerequisites

Upstream Artifact Verification (CRITICAL)

Before creating this document, you MUST:

1. List existing upstream artifacts: ls docs/01_BRD/ docs/02_PRD/ docs/03_EARS/ docs/04_BDD/ docs/05_ADR/ docs/06_SYS/ docs/07_REQ/ 2>/dev/null
2. Reference only existing documents in traceability tags
3. Use null only when upstream artifact type genuinely doesn't exist
4. NEVER use placeholders like BRD-XXX or TBD
5. Do NOT create missing upstream artifacts - skip functionality instead

Before creating a BRD, read:

1. Shared Standards: .claude/skills/doc-flow/SHARED_CONTENT.md
2. Template: ai_dev_ssd_flow/01_BRD/BRD-MVP-TEMPLATE.md
3. Creation Rules: ai_dev_ssd_flow/01_BRD/BRD-MVP-TEMPLATE.md
4. Validation Rules: ai_dev_ssd_flow/01_BRD/BRD_MVP_SCHEMA.yaml
5. Platform vs Feature Guide: ai_dev_ssd_flow/PLATFORM_VS_FEATURE_BRD.md

For New Projects: Use project-init skill first to initialize project structure.

Pre-Flight Check (MANDATORY)

Before creating ANY BRD section, confirm:

1. ✅ Read ai_dev_ssd_flow/ID_NAMING_STANDARDS.md - Element Type Codes table
2. ✅ Element ID format: BRD.{DOC_NUM}.{HASH} (3 segments, dots)

Common Element Types:

⚠️ Removed Patterns: Do NOT use AC-XXX, FR-XXX, BC-XXX, BO-XXX formats.

When to Use This Skill

Use doc-brd when:

• Starting a new project or feature
• Defining business requirements and objectives
• Documenting strategic alignment and market context
• Establishing success criteria and stakeholder needs
• Translating implementation plans into business-language BRD sections
• Refining BRDs generated from small reference documents or prompt-first workflows
• You are at Layer 1 of the SDD workflow

BRD Categorization: Platform vs Feature

CRITICAL DECISION: Before creating a BRD, determine if it's a Platform BRD or Feature BRD.

Questionnaire

1. Does this BRD define infrastructure, technology stack, or cross-cutting concerns?

- Yes → Likely Platform BRD - No → Continue

1. Does this BRD describe a specific user-facing workflow or feature?

- Yes → Likely Feature BRD - No → Continue

1. Will other BRDs depend on or reference this BRD's architectural decisions?

- Yes → Likely Platform BRD - No → Likely Feature BRD

1. Does this BRD establish patterns, standards, or capabilities used across multiple features?

- Yes → Platform BRD - No → Feature BRD

1. Does this BRD implement functionality on top of existing platform capabilities?

- Yes → Feature BRD - No → Platform BRD

Auto-Detection Logic

• Title contains "Platform", "Architecture", "Infrastructure", "Integration" → Platform BRD
• Title contains specific workflow names, user types (B2C, B2B), or feature names → Feature BRD
• References/depends on BRD-01 or foundation ADRs → Feature BRD
• Establishes technology choices or system-wide patterns → Platform BRD

Workflow Differences

Platform BRD Path:

1. Create Platform BRD (populate sections 3.6 and 3.7)
2. Create ADRs for critical technology decisions (identified in BRD sections 3.6/3.7)
3. Create PRD referencing Platform BRD and ADRs
4. Create additional ADRs for implementation details
5. Continue to SPEC

Feature BRD Path:

1. Create Feature BRD (reference Platform BRD in sections 3.6 and 3.7)
2. Create PRD for feature
3. Create ADRs for implementation decisions (if needed)
4. Continue to SPEC

Section 3.6 & 3.7 Rules

Platform BRD:

• MUST populate Section 3.6 (Technology Stack Prerequisites) with detailed technology choices
• MUST populate Section 3.7 (Mandatory Technology Conditions) with non-negotiable constraints

Feature BRD:

• MUST mark Section 3.6 as "N/A - See Platform BRD-NN Section 3.6" and reference specific items
• MUST mark Section 3.7 as "N/A - See Platform BRD-NN Section 3.7" and reference specific conditions

Reference: ai_dev_ssd_flow/PLATFORM_VS_FEATURE_BRD.md for detailed guidance

BRD-Specific Guidance

1. Template Selection

Primary Template:

BRD-MVP-TEMPLATE.md - Comprehensive business requirements (general purpose)

• Use for: All business requirements documents
• Sections: Complete 18-section structure
• Best for: Complex projects, regulatory compliance needs
• Location: ai_dev_ssd_flow/01_BRD/BRD-MVP-TEMPLATE.md

Note: Use the comprehensive template for all BRD documents. For simpler requirements, complete only the essential sections and mark others as "N/A - Not applicable for this scope".

Note: Technical QA standards, testing strategy, and defect management are documented in PRD-TEMPLATE.md Section 21 (product level).

Nested Folder Rule (MANDATORY): ALL BRDs MUST be in nested folders regardless of document size.

Monolithic Structure - for MVP/small documents:

• Location: docs/01_BRD/BRD-NN_{slug}/BRD-NN_{slug}.md (INSIDE nested folder)
• H1 Title: # BRD-NN: Document Title (no .S suffix)
• Template: ai_dev_ssd_flow/01_BRD/BRD-MVP-TEMPLATE.md
• CRITICAL: Even monolithic BRDs MUST be in a nested folder

Section-Based Structure - for large/complex documents:

• Location: docs/01_BRD/BRD-NN_{slug}/BRD-NN.S_{section}.md
• Folder Naming: BRD-NN_{slug}/ where slug MUST match the index file slug
• H1 Title: # BRD-NN.S: Section Title (includes .S suffix)
• Index template: ai_dev_ssd_flow/01_BRD/BRD-SECTION-0-TEMPLATE.md
• Content template: ai_dev_ssd_flow/01_BRD/BRD-SECTION-TEMPLATE.md
• Reference: ai_dev_ssd_flow/ID_NAMING_STANDARDS.md (Section-Based File Splitting)

2. Required Sections (18 Total)

Document Control (MANDATORY - First section before all numbered sections):

• Project Name
• Document Version
• Date (YYYY-MM-DD)
• Document Owner
• Prepared By
• Status (Draft, In Review, Approved, Superseded)
• Document Revision History table

Core Sections (18-Section Structure):

1. Introduction - Purpose, scope, audience
2. Business Objectives - Goals, hypothesis, metrics
3. Project Scope - Boundaries, workflows
4. Stakeholders - Decision makers
5. User Stories - High-level needs
6. Functional Requirements - Business capabilities
7. Quality Attributes - Performance, security, ADR Topics (Section 7.2)
8. Business Constraints and Assumptions - Limitations
9. Acceptance Criteria - Success measures
10. Business Risk Management - Risk register
11. Implementation Approach - Phases, rollout
12. Support and Maintenance - Support model
13. Cost-Benefit Analysis - ROI, costs
14. Project Governance - Decision authority, approval (14.5)
15. Quality Assurance - QA standards, testing strategy
16. Traceability - Requirements matrix, health score
17. Glossary - Terms (6 subsections: 17.1-17.6)
18. Appendices - Supporting materials and references

Platform BRD Additional Sections:

• 3.6 Technology Stack Prerequisites (MUST populate for Platform BRD)
• 3.7 Mandatory Technology Conditions (MUST populate for Platform BRD)

3. Upstream Source Configuration

BRDs can be created with or without upstream reference documents.

Default: No Upstream Sources

Most BRDs are created from stakeholder interviews, product ideas, or business requirements without formal source documents. Use the default configuration:

custom_fields:
  upstream_mode: "none"  # Default - no drift detection

With Reference Documents

If BRD content is derived from documents in docs/00_REF/, configure tracking:

custom_fields:
  upstream_mode: "ref"
  upstream_ref_path: "../../00_REF/source_docs/"  # Relative to BRD location

With Implementation Plans (IPLAN-*)

If BRD content is primarily derived from an implementation plan, keep drift mode default unless reference documents are primary:

custom_fields:
  upstream_mode: "none"

When implementation plan input is supplemented by reference documents, set upstream_mode: "ref" and track only the reference path(s) in upstream_ref_path.

Multiple Reference Folders

For nested or multiple reference folders:

custom_fields:
  upstream_mode: "ref"
  upstream_ref_path:
    - "../../00_REF/source_docs/"
    - "../../00_REF/foundation/"
    - "../../00_REF/internal_ops/cloud_cost_monitoring/"

Path Resolution

Paths are relative to the BRD file location:

• BRD at: docs/01_BRD/BRD-01_platform/BRD-01.md
• REF at: docs/00_REF/source_docs/
• Path: "../../00_REF/source_docs/"

Behavior Summary

Input Mode Authoring Guardrail

For plan-derived BRDs:

• Convert implementation-heavy details into business outcomes, risks, and acceptance criteria.
• Keep low-level execution mechanics in the IPLAN, not in BRD functional requirements.
• If objective/scope in source inputs conflict, resolve before writing BRD sections 1-3.

4. Architecture Decision Requirements Section (7.2) - MANDATORY

Purpose: Identify architectural topics requiring decisions with cost-focused, alternatives-based analysis.

Every BRD MUST include Section 7.2: "Architecture Decision Requirements" addressing all 7 mandatory ADR topic categories.

7 Mandatory ADR Topic Categories

Element Type Code: 32 = Architecture Topic

Required Fields Per Topic

Alternatives Overview Table (MANDATORY for Selected status)

**Alternatives Overview**:

| Option | Function | Est. Monthly Cost | Selection Rationale |
|--------|----------|-------------------|---------------------|
| Serverless | Event-driven compute | $200-$800 | Selected - cost-effective |
| Kubernetes | Container orchestration | $500-$2,000 | Rejected - over-engineered |
| VM-based | Traditional VMs | $400-$1,500 | Rejected - manual scaling |

Cloud Provider Comparison Table (MANDATORY for Selected status)

**Cloud Provider Comparison**:

| Criterion | GCP | Azure | AWS |
|-----------|-----|-------|-----|
| **Service Name** | Cloud Run | Container Apps | Fargate |
| **Est. Monthly Cost** | $300 | $350 | $400 |
| **Key Strength** | Auto-scaling | AD integration | Ecosystem |
| **Key Limitation** | Fewer features | Higher cost | Complex pricing |
| **Fit for This Project** | High | Medium | Medium |

Complete Topic Example

### BRD.01.3201: Infrastructure

**Status**: Selected

**Business Driver**: Customer onboarding workflow requires scalable compute for variable registration volumes.

**Business Constraints**:
- Must support auto-scaling for peak periods (10x baseline)
- Maximum infrastructure cost: $5,000/month
- 99.9% availability during business hours

**Alternatives Overview**:

| Option | Function | Est. Monthly Cost | Selection Rationale |
|--------|----------|-------------------|---------------------|
| Serverless (Cloud Functions) | Event-driven compute | $200-$800 | Selected - cost-effective for variable load |
| Kubernetes (GKE/EKS) | Container orchestration | $500-$2,000 | Rejected - over-engineered for this scale |
| VM-based (Compute Engine) | Traditional VMs | $400-$1,500 | Rejected - manual scaling overhead |

**Cloud Provider Comparison**:

| Criterion | GCP | Azure | AWS |
|-----------|-----|-------|-----|
| **Service Name** | Cloud Run | Container Apps | Lambda + Fargate |
| **Est. Monthly Cost** | $300 | $350 | $400 |
| **Key Strength** | Simple container deployment | Azure AD integration | Largest ecosystem |
| **Key Limitation** | Fewer enterprise features | Higher baseline cost | Complex pricing |
| **Fit for This Project** | High | Medium | Medium |

**Recommended Selection**: GCP Cloud Run - serverless containers with optimal cost-to-scale ratio.

**PRD Requirements**: Evaluate cold start impact on latency. Specify concurrency limits and scaling policies.

Handling N/A and Pending Status

N/A Example:

### BRD.01.3206: AI/ML Architecture

**Status**: N/A - No AI/ML components in project scope

**Reason**: This feature uses standard rule-based validation. No machine learning, AI agents, or predictive analytics required.

**PRD Requirements**: None for current scope. Flag for Phase 2 evaluation if ML-based fraud detection needed.

Pending Example:

### BRD.01.3205: Observability

**Status**: Pending - Awaiting infrastructure finalization

**Business Driver**: System monitoring required for SLA compliance.

**Business Constraints**:
- Real-time alerting for system failures
- Minimum 30-day log retention

**Alternatives Overview**: [Placeholder - To be completed after infrastructure selection]

**PRD Requirements**: Complete observability analysis after infrastructure finalization.

Layer Separation Principle

BRD Section 7.2          →    PRD Section 18         →    ADR
(WHAT & WHY & HOW MUCH)       (HOW to evaluate)          (Final decision)
─────────────────────────────────────────────────────────────────────────
Business drivers              Technical details          Implementation decision
Business constraints          Deep-dive analysis         Trade-off analysis
Cost estimates                Evaluation criteria        Selected approach

Do NOT write: "See ADR-033" or "Reference ADR-045" (ADRs don't exist yet)

Reference: See ai_dev_ssd_flow/01_BRD/BRD-MVP-TEMPLATE.md Section 9 for complete guidelines

5. Document Control Section Positioning

CRITICAL: Document Control MUST be the first section at the very top of the BRD (before all numbered sections).

Correct Structure:

# BRD-01: Project Name

## Document Control
[All metadata fields here]

## 1. Executive Summary
[Content here]

Cumulative Tagging Requirements

Layer 1 (BRD): No upstream tags required (entry point)

Tag Count: 0 tags

Format: BRD has no @ tags since it's Layer 1 (top of hierarchy)

Downstream artifacts will tag BRD (using unified format):

• PRD will include: @brd: BRD.01.0130 (TYPE.NN.xxxx format)
• EARS will include: @brd: BRD.01.0130
• All downstream artifacts inherit BRD tags

Tag Format Convention (By Design)

The SDD framework uses two distinct notation systems for cross-references:

Key Distinction:

• @adr: ADR-033 → Points to the document ADR-033_risk_limit_enforcement.md
• @brd: BRD.17.0101 → Points to element 01.01 inside document BRD-017.md

Unified Element ID Format (MANDATORY)

For hierarchical requirements (BRD, PRD, EARS, BDD, SYS, REQ):

• Always use: TYPE.NN.xxxx (dot separator, 3-segment unified format)
• Never use: TYPE-NN:NNN (colon separator - DEPRECATED)
• Never use: TYPE.NN.TT.SS (old 4-segment format - DEPRECATED)

Examples:

• @brd: BRD.17.0101 ✅
• @brd: BRD.017.001 ❌ (old 3-segment format)

Upstream/Downstream Artifacts

Upstream Sources (what drives BRD creation):

• Reference documents (docs/00_REF/) - if upstream_mode: "ref"
• Business owner requirements (verbal/written)
• Stakeholder interviews
• Market analysis
• Product ideas

Downstream Artifacts (what BRD drives):

• PRD (Layer 2) - Product requirements derived from BRD
• EARS (Layer 3) - Formal requirements from BRD business needs
• BDD (Layer 4) - Test scenarios validating BRD objectives
• ADR (Layer 5) - Architecture decisions for topics identified in BRD Section "Architecture Decision Requirements"

Same-Type Document Relationships (conditional):

• @related-brd: BRD-NN - BRDs sharing business domain context
• @depends-brd: BRD-NN - BRD that must be implemented first (e.g., platform BRD before feature BRD)

Creation Process

Step 1: Determine BRD Type

Use questionnaire above to determine Platform vs Feature BRD.

Step 2: Read Strategy Documents

Read relevant {project_root}/strategy/ sections to understand business logic.

Step 3: Select Template

Choose appropriate template (comprehensive, simplified, or domain-specific).

Step 4: Reserve ID Number

Check docs/01_BRD/ for next available ID number (e.g., BRD-01, BRD-02).

ID Numbering Convention: Start with 2 digits and expand only as needed.

• ✅ Correct: BRD-01, BRD-99, BRD-102
• ❌ Incorrect: BRD-001, BRD-009 (extra leading zero not required)

Step 5: Create BRD Folder and Files

Folder structure (DEFAULT - nested folder per document with descriptive slug):

1. Create folder: docs/01_BRD/BRD-NN_{slug}/ (folder slug MUST match index file slug)
2. Create index file: docs/01_BRD/BRD-NN_{slug}/BRD-NN.0_{section_type}.md (shortened, PREFERRED)
3. Create section files: docs/01_BRD/BRD-NN_{slug}/BRD-NN.S_{section_type}.md (shortened, PREFERRED)

Example (Shortened Pattern - PREFERRED):

docs/01_BRD/BRD-01_platform_architecture/
├── BRD-01.md
├── BRD-01.1_executive_summary.md
├── BRD-01.2_business_context.md
└── BRD-01.3_requirements.md

Note: Folder contains descriptive slug, so filenames can omit it. Full pattern (BRD-01.0_platform_architecture_index.md) also accepted for backward compatibility.

Monolithic Option (for small documents ≤25KB): docs/01_BRD/BRD-NN_{slug}/BRD-NN_{slug}.md (still in nested folder)

Step 6: Fill Document Control Section

Complete all required metadata fields and initialize Document Revision History table.

Step 7: Complete Core Sections

Fill all 18 required sections following template structure.

Platform BRD: Populate sections 3.6 and 3.7 with technology details Feature BRD: Mark sections 3.6 and 3.7 as "N/A - See Platform BRD-NN"

Step 8: Document Architecture Decision Requirements

List topics needing architectural decisions (do NOT reference specific ADR numbers).

Step 9: Configure Upstream Sources

If BRD is derived from reference documents in docs/00_REF/:

custom_fields:
  upstream_mode: "ref"
  upstream_ref_path: "../../00_REF/source_docs/"

If BRD created from scratch (stakeholder input, product ideas):

custom_fields:
  upstream_mode: "none"  # Default - drift detection skipped

Step 10: Create/Update Traceability Matrix

MANDATORY: Create or update docs/01_BRD/BRD-00_TRACEABILITY_MATRIX.md

• Use template: ai_dev_ssd_flow/01_BRD/BRD-00_TRACEABILITY_MATRIX-TEMPLATE.md
• Add BRD entry with upstream sources and downstream artifacts
• Update traceability matrix in same commit after BRD validation passes (see SHARED_CONTENT.md Traceability Matrix Update Workflow)

Step 11: Validate BRD

Run unified BRD validation wrapper (primary gate):

# Core blocking checks (used by pre-commit and CI)
bash ai_dev_ssd_flow/01_BRD/scripts/validate_brd_wrapper.sh docs/01_BRD --skip-advisory

# Optional full tiered run (includes advisory checks)
bash ai_dev_ssd_flow/01_BRD/scripts/validate_brd_wrapper.sh docs/01_BRD

Step 12: Commit Changes

Commit BRD file and traceability matrix together.

Validation

Automated Validation

BRD-Specific Validation:

bash ai_dev_ssd_flow/01_BRD/scripts/validate_brd_wrapper.sh docs/01_BRD --skip-advisory

Quality Gates Validation:

./scripts/validate_quality_gates.sh docs/01_BRD/BRD-01_platform.md

Manual Checklist

• Document Control section at top (before all numbered sections)
• All required metadata fields completed
• Document Revision History table initialized
• BRD type determined (Platform vs Feature)
• Sections 3.6 & 3.7 handled correctly for BRD type
• Architecture Decision Requirements listed (no ADR numbers referenced)
• Strategy references in Traceability section
• All 18 sections completed
• Traceability matrix created/updated
• No broken links
• File size <50,000 tokens (standard) or <100,000 tokens (maximum)

Diagram Standards

All diagrams MUST use Mermaid syntax. Text-based diagrams (ASCII art, box drawings) are prohibited. See: ai_dev_ssd_flow/DIAGRAM_STANDARDS.md and mermaid-gen skill.

BRD Diagram Contract (MANDATORY):

• Include @diagram: c4-l1 and @diagram: dfd-l0
• If sequence diagram exists, include one sequence contract tag (@diagram: sequence-sync|sequence-async|sequence-error)
• Add intent header fields above each required diagram block: diagram_type, level, scope_boundary, upstream_refs, downstream_refs

Common Pitfalls

1. Referencing ADR numbers: Don't write "See ADR-033" in Architecture Decision Requirements section (ADRs don't exist yet)
2. Wrong sections 3.6/3.7 treatment: Platform BRD must populate, Feature BRD must reference Platform BRD
3. Wrong upstream_mode: Set upstream_mode: "ref" only if deriving from docs/00_REF/ documents
4. Document Control not first: Must be at very top before all numbered sections
5. Skipping traceability matrix: MANDATORY to create/update matrix

Post-Creation Validation (MANDATORY - NO CONFIRMATION)

CRITICAL: Execute this validation loop IMMEDIATELY after document creation. Do NOT proceed to next document until validation passes.

Automatic Validation Loop

LOOP:
  1. Run BRD template validation script
  2. IF errors found: Fix issues
  3. IF warnings found: Review and address
  4. IF unfixable issues: Log for manual review, continue
  5. IF clean: Mark VALIDATED, proceed

Validation Command

# Unified BRD core validation (primary)
bash ai_dev_ssd_flow/01_BRD/scripts/validate_brd_wrapper.sh docs/01_BRD --skip-advisory

# Unified BRD full tiered validation (optional)
bash ai_dev_ssd_flow/01_BRD/scripts/validate_brd_wrapper.sh docs/01_BRD

# Component-level diagnostics (secondary, optional)
python ai_dev_ssd_flow/01_BRD/scripts/validate_brd.py docs/01_BRD/BRD-NN_slug.md

Layer-Specific Upstream Requirements

Auto-Fix Actions (No Confirmation Required)

Validation Codes Reference

Quality Gate

Blocking: YES - Cannot proceed to next document until Phase 1 validation passes with 0 errors.

Next Skill

After creating BRD, use:

doc-prd - Create Product Requirements Document (Layer 2)

The PRD will:

• Reference this BRD as upstream source
• Include @brd: BRD.NN.01.SS tags (unified 3-segment format)
• Define product features and KPIs
• Inherit Architecture Decision Requirements topics

Reference Documents

For supplementary documentation related to BRD artifacts:

• Format: BRD-REF-NNN_{slug}.md
• Skill: Use doc-ref skill
• Validation: Reduced (4 checks only)
• Examples: Project overviews, executive summaries, stakeholder guides

BRD-REF Ready-Score Exemption

BRD-REF documents are EXEMPT from ready-scores and quality gates:

Purpose: BRD-REF documents are reference targets that other documents link to. They provide supporting information, context, or external references but do not define formal business requirements.

Reference: See ai_dev_ssd_flow/01_BRD/BRD_MVP_SCHEMA.yaml for validation details.

Related Resources

• Main Guide: ai_dev_ssd_flow/SPEC_DRIVEN_DEVELOPMENT_GUIDE.md
• Platform vs Feature Guide: ai_dev_ssd_flow/PLATFORM_VS_FEATURE_BRD.md
• BRD Creation Rules: ai_dev_ssd_flow/01_BRD/BRD-MVP-TEMPLATE.md
• BRD Validation Rules: ai_dev_ssd_flow/01_BRD/BRD_MVP_SCHEMA.yaml
• BRD README: ai_dev_ssd_flow/01_BRD/README.md
• Shared Standards: .claude/skills/doc-flow/SHARED_CONTENT.md
• BRD Audit Skill: .claude/skills/doc-brd-audit/SKILL.md (unified; deprecated: doc-brd-validator, doc-brd-reviewer)
• Naming Standards Skill: .claude/skills/doc-naming/SKILL.md

Quick Reference

BRD Purpose: Define business needs and objectives

Layer: 1 (Entry point)

Tags Required: None (0 tags)

Key Decision: Platform vs Feature BRD

Critical Sections:

• 3.6 Technology Stack Prerequisites (Platform BRD populates, Feature BRD references)
• 3.7 Mandatory Technology Conditions (Platform BRD populates, Feature BRD references)
• Architecture Decision Requirements (list topics, NOT ADR numbers)

Next: doc-prd

Version History