doc-bdd-autopilot（文档 bdd 自动驾驶仪）

doc-bdd-autopilot

Purpose

Automated BDD (Behavior-Driven Development) generation pipeline that processes EARS documents and generates Gherkin scenarios with readiness validation, scenario categorization, and ADR-Ready score compliance.

Layer: 4 (BDD Generation)

Upstream: BRD (Layer 1), PRD (Layer 2), EARS (Layer 3)

Downstream Artifacts: ADR (Layer 5), SYS (Layer 6), REQ (Layer 7)

Input Contract (IPLAN-004 Standard)

• Supported modes:

- --ref <path> - --prompt "<text>" - --iplan <path|IPLAN-NNN>

• Precedence: --iplan > --ref > --prompt
• IPLAN resolution order:

1. Use explicit file path when it exists 2. Resolve work_plans/IPLAN-NNN*.md 3. Resolve governance/plans/IPLAN-NNN*.md 4. If multiple matches exist, fail with disambiguation request

• Merge conflict rule:

- Objective/scope conflicts between primary and supplemental sources are blocking and require user clarification.

Input Contract (IPLAN-004 Standard)

• Supported modes:

- --ref <path> - --prompt "<text>" - --iplan <path|IPLAN-NNN>

• Precedence: --iplan > --ref > --prompt
• IPLAN resolution order:

1. Use explicit file path when it exists 2. Resolve work_plans/IPLAN-NNN*.md 3. Resolve governance/plans/IPLAN-NNN*.md 4. If multiple matches exist, fail with disambiguation request

• Merge conflict rule:

- Objective/scope conflicts between primary and supplemental sources are blocking and require user clarification.

Skill Dependencies

This autopilot orchestrates the following skills:

Delegation Principle: The autopilot orchestrates workflow but delegates:

• BDD structure/content rules -> doc-bdd skill
• Real-time quality feedback -> quality-advisor skill
• BDD validation logic -> doc-bdd-validator skill
• Content review and scoring -> doc-bdd-reviewer skill
• Issue resolution and fixes -> doc-bdd-fixer skill
• EARS validation logic -> doc-ears-validator skill
• Element ID standards -> doc-naming skill

Document Type Contract (MANDATORY)

When generating BDD document instances, the autopilot MUST:

1. Read instance_document_type from template:

- Source: ai_dev_ssd_flow/04_BDD/BDD-MVP-TEMPLATE.feature - Field: N/A (Gherkin files use header comments)

1. Set document_type in generated document:

- BDD uses Gherkin .feature format without YAML frontmatter - Add header comment: # document_type: bdd-document - Schema validates via BDD_MVP_SCHEMA.yaml

1. Validation: Generated documents MUST indicate document_type: bdd-document

- Templates have header comment # document_type: template - Instances have header comment # document_type: bdd-document

Error Handling: If header comment is missing from template, add # document_type: bdd-document.

Smart Document Detection

The autopilot automatically determines the action based on the input document type.

Input Type Recognition

Detection Algorithm

1. Parse input: Extract TYPE and NN from "{TYPE}-{NN}"
2. Determine action:
   - IF TYPE == "BDD": Review Mode
   - ELSE IF TYPE == "EARS": Generate/Find Mode
   - ELSE: Error (invalid type for this autopilot)
3. For Generate/Find Mode:
   - Check: Does BDD-{NN} exist in docs/04_BDD/?
   - IF exists: Switch to Review Mode for BDD-{NN}
   - ELSE: Proceed with Generation from EARS-{NN}

File Existence Check

# Check for nested folder structure (mandatory)
ls docs/04_BDD/BDD-{NN}_*/

Examples

# Review mode (same type - BDD input)
/doc-bdd-autopilot BDD-01           # Reviews existing BDD-01

# Generate/Find mode (upstream type - EARS input)
/doc-bdd-autopilot EARS-01          # Generates BDD-01 if missing, or reviews existing BDD-01

# Multiple inputs
/doc-bdd-autopilot EARS-01,EARS-02  # Generates/reviews BDD-01 and BDD-02
/doc-bdd-autopilot BDD-01,BDD-02    # Reviews BDD-01 and BDD-02

Action Determination Output

Input: EARS-01
├── Detected Type: EARS (upstream)
├── Expected BDD: BDD-01
├── BDD Exists: Yes → docs/04_BDD/BDD-01_f1_iam/
└── Action: REVIEW MODE - Running doc-bdd-reviewer on BDD-01

Input: EARS-05
├── Detected Type: EARS (upstream)
├── Expected BDD: BDD-05
├── BDD Exists: No
└── Action: GENERATE MODE - Creating BDD-05 from EARS-05

Input: BDD-03
├── Detected Type: BDD (self)
└── Action: REVIEW MODE - Running doc-bdd-reviewer on BDD-03

When to Use This Skill

Use doc-bdd-autopilot when:

• You have one or more completed EARS documents ready for BDD generation
• You want automated scenario categorization (Success, Error, Edge, Data-Driven)
• You need BDD-Ready score validation before generation
• You want automatic ADR-Ready score validation after BDD creation
• You need to generate Gherkin Given-When-Then scenarios from EARS statements

Do NOT use when:

• Creating a single BDD suite with extensive manual customization (use doc-bdd directly)
• EARS documents are incomplete or in Draft status
• You need fine-grained control over each scenario
• EARS BDD-Ready score is below 90%

Workflow Overview

flowchart TD
    subgraph Phase1["Phase 1: EARS Analysis"]
        A[Input: EARS List] --> B[Read EARS Documents]
        B --> C[Extract EARS Statements]
        C --> D[Categorize by Statement Type]
        D --> E[Identify Quality Attributes]
        E --> F[Extract Threshold References]
    end

    subgraph Phase2["Phase 2: BDD Readiness Check"]
        F --> G{Next EARS}
        G --> H[Check BDD-Ready Score]
        H --> I{Score >= 90?}
        I -->|No| J[Auto-Fix EARS Issues]
        J --> K[Re-validate EARS]
        K --> I
        I -->|Yes| L[Mark EARS Ready]
    end

    subgraph Phase3["Phase 3: BDD Generation"]
        L --> M[Plan Section Structure]
        M --> N[Generate Feature Files]
        N --> O[Generate Success Path Scenarios]
        O --> P[Generate Error Condition Scenarios]
        P --> Q[Generate Edge Case Scenarios]
        Q --> R[Generate Data-Driven Scenarios]
        R --> S[quality-advisor: Real-time Feedback]
        S --> T[Add Quality Attribute Scenarios]
        T --> U[Add Cumulative Tags]
        U --> V[Write BDD Files]
    end

    subgraph Phase4["Phase 4: BDD Validation"]
        V --> W[Run doc-bdd-validator]
        W --> X{ADR-Ready >= 90?}
        X -->|No| Y[Auto-Fix BDD Issues]
        Y --> Z[Re-validate BDD]
        Z --> X
        X -->|Yes| AA[Validation Passed]
    end

    subgraph Phase5["Phase 5: Final Review"]
        AA --> AB[Check Gherkin Syntax Compliance]
        AB --> AC[Verify Threshold References]
        AC --> AD[Validate Cumulative Tags]
        AD --> AE[Check Scenario Coverage]
        AE --> AF{Review Passed?}
        AF -->|No| AG[Flag Issues]
        AG --> AH[Auto-Fix or Manual]
        AH --> AF
        AF -->|Yes| AI[Mark BDD Complete]
    end

    AI --> AJ{More EARS?}
    AJ -->|Yes| G
    AJ -->|No| AK[Generate Summary Report]
    AK --> AL[Complete]

Detailed Workflow

Phase 1: EARS Analysis

Analyze EARS documents to extract requirements for BDD scenario generation.

Input Sources (from EARS):

Analysis Process:

# Check for EARS documents
ls -la docs/03_EARS/

# Expected structure:
# - EARS-NN_{slug}.md (monolithic)
# - EARS-NN_{slug}/ (sectioned with EARS-NN.S_{section}.md files)

Output: Scenario catalog with categorization hints, threshold references, and traceability links.

Phase 2: BDD Readiness Check

Validate that source EARS meet BDD-Ready requirements before generation.

Skill Delegation: This phase uses validation rules from doc-ears-validator skill. See: .claude/skills/doc-ears-validator/SKILL.md for complete EARS validation rules.

BDD-Ready Scoring Criteria (100%):

Minimum Score: 90% (configurable)

Auto-Fix Actions:

Validation Command (internal):

python ai_dev_ssd_flow/04_BDD/scripts/validate_bdd_ready.py \
  --ears docs/03_EARS/EARS-01_{slug}/ \
  --min-score 90 \
  --auto-fix

Phase 3: BDD Generation

Generate BDD scenarios from validated EARS with real-time quality feedback.

Skill Delegation: This phase follows rules defined in doc-bdd skill. See: .claude/skills/doc-bdd/SKILL.md for complete BDD creation guidance. Quality Guidance: Uses quality-advisor skill for real-time feedback during generation. See: .claude/skills/quality-advisor/SKILL.md for quality monitoring.

Generation Process:

1. Reserve BDD ID: # Check for next available ID ls docs/04_BDD/BDD-*.feature docs/04_BDD/BDD-*/BDD-*.0_*.md 2>/dev/null | \ grep -oP 'BDD-\K\d+' | sort -n | tail -1 # Increment for new BDD
2. Create Suite Directory: mkdir -p docs/04_BDD/BDD-NN_{slug}/
3. Plan Section Structure: Section Planning Based on EARS Scope: EARS Statement Count BDD Structure Rationale 1-10 statements 2-3 sections Compact coverage 11-30 statements 4-8 sections Standard grouping 31+ statements 9+ sections with subsections Complex domain
4. Generate Index File (BDD-NN.0_index.md): Field Value BDD ID BDD-NN Version 0.1.0 Date Created Current date (YYYY-MM-DD) Last Updated Current date (YYYY-MM-DD) Status Draft Priority From EARS priority Source Document @ears: EARS.NN.25.SS (single value) ADR-Ready Score Calculated after generation
5. Generate Feature Files by Section: Section File Pattern: BDD-NN.S_{section_name}.feature Subsection Pattern (if >800 lines): BDD-NN.S.SS_{subsection_name}.feature
6. Generate Scenarios by Category: Success Path Scenarios (@primary): @section: NN.SS @parent_doc: BDD-NN @index: BDD-NN.0_index.md @brd:BRD.NN.01.SS @prd:PRD.NN.07.SS @ears:EARS.NN.25.SS Feature: BDD-NN.SS: [Feature Name] As a [role] I want [feature] So that [benefit] Background: Given the system timezone is "America/New_York" And the current time is "09:30:00" in "America/New_York" @primary @functional @scenario-id:BDD.NN.14.01 Scenario: Successful [action description] Given [precondition from EARS WHEN clause] When [action from EARS trigger] Then [outcome from EARS SHALL clause] And response time is less than @threshold:PRD.NN.perf.api.p95_latency Error Condition Scenarios (@negative): @negative @error_handling @scenario-id:BDD.NN.14.10 Scenario: [Error condition] results in [expected behavior] Given [error precondition from EARS IF clause] When [action that triggers error] Then [error handling from EARS prevention/recovery] And error code "[ERROR_CODE]" is returned Edge Case Scenarios (@edge_case, @boundary): @edge_case @boundary @scenario-id:BDD.NN.14.20 Scenario: [Boundary condition] at [limit value] Given [boundary precondition] When [action at boundary] Then [expected boundary behavior] Data-Driven Scenarios (@data_driven): @data_driven @scenario-id:BDD.NN.14.30 Scenario Outline: [Parameterized test description] Given [context with <variable>] When [action with <variable>] Then [outcome with <variable>] Examples: | variable | expected | | value1 | result1 | | value2 | result2 | Quality Attribute Scenarios (@quality_attribute): @quality_attribute @performance @scenario-id:BDD.NN.14.40 Scenario: API response meets performance threshold Given system is under normal load When user submits request Then response time is less than @threshold:PRD.NN.perf.api.p95_latency And throughput exceeds @threshold:PRD.NN.perf.api.min_throughput Integration Scenarios (@integration): @integration @external_system @scenario-id:BDD.NN.14.50 Scenario: External system integration succeeds Given external service is available When system initiates integration call Then integration completes within @threshold:PRD.NN.timeout.integration.max Failure Recovery Scenarios (@failure_recovery): @failure_recovery @circuit_breaker @scenario-id:BDD.NN.14.60 Scenario: System recovers from transient failure Given external service experiences transient failure When retry mechanism activates Then system recovers within @threshold:PRD.NN.retry.max_attempts attempts And circuit breaker state is updated
7. Real-Time Quality Feedback (via quality-advisor skill):

- Monitor Gherkin syntax compliance as scenarios are generated - Detect anti-patterns (missing Given/When/Then, hardcoded values) - Validate @threshold tag format and references - Check element ID format compliance (BDD.NN.14.SS, BDD.NN.15.SS) - Flag issues early to reduce post-generation rework

1. Add Cumulative Traceability Tags: Required Tags per Feature (Gherkin-native, NOT in comments): @brd:BRD.NN.01.SS @prd:PRD.NN.07.SS @ears:EARS.NN.25.SS Feature: BDD-NN.SS: Feature Name Layer 4 Cumulative Tag Requirement: @brd, @prd, @ears (3+ tags)
2. Generate Redirect Stub: # Create redirect stub at docs/04_BDD/ root touch docs/04_BDD/BDD-NN_{slug}.feature Redirect Stub Content: @redirect @section: NN.00 @parent_doc: BDD-NN @index: BDD-NN.0_index.md Feature: BDD-NN: [Suite Name] (Redirect) This is a redirect stub. Test scenarios are in section files: - BDD-NN.1_{section1}.feature - [Description] - BDD-NN.2_{section2}.feature - [Description] Background: Given the system timezone is "America/New_York" # No scenarios in redirect stub
3. Update Index File:

- List all section files with scenario counts and line counts - Add traceability matrix linking EARS to BDD sections - Calculate and display ADR-Ready Score

Phase 4: BDD Validation

After BDD generation, validate structure and ADR-Ready score.

Skill Delegation: This phase uses validation rules from doc-bdd-validator skill. See: .claude/skills/doc-bdd-validator/SKILL.md for complete validation rules.

Validation Command:

python ai_dev_ssd_flow/04_BDD/scripts/validate_bdd.py docs/04_BDD/BDD-NN_{slug}/ --verbose

Validation Checks:

Auto-Fix Actions:

Validation Loop:

LOOP (max 3 iterations):
  1. Run doc-bdd-validator
  2. IF errors found: Apply auto-fixes
  3. IF warnings found: Review and address if critical
  4. IF ADR-Ready Score < 90%: Enhance scenarios
  5. IF clean: Mark VALIDATED, proceed
  6. IF max iterations: Log issues, flag for manual review

Phase 5: Review & Fix Cycle (v2.1)

Iterative review and fix cycle to ensure BDD quality before completion.

flowchart TD
    A[Phase 5 Start] --> B[Run doc-bdd-reviewer]
    B --> C[Generate Review Report]
    C --> D{Review Score >= 90?}

    D -->|Yes| E[PASS - Proceed to Phase 6]
    D -->|No| F{Iteration < Max?}

    F -->|Yes| G[Run doc-bdd-fixer]
    G --> H[Apply Fixes]
    H --> I[Generate Fix Report]
    I --> J[Increment Iteration]
    J --> B

    F -->|No| K[Flag for Manual Review]
    K --> L[Generate Final Report with Remaining Issues]
    L --> E

5.1 Initial Review

Run doc-bdd-reviewer to identify issues.

/doc-bdd-reviewer BDD-NN

Output: BDD-NN.A_audit_report_v001.md (legacy-compatible reviewer report may still exist)

Review Checks:

1. Gherkin Syntax Compliance:

- All scenarios use Given-When-Then structure - No vague language ("fast", "efficient", "user-friendly") - All steps are atomic and actionable - Background steps are appropriate

1. Threshold Reference Consistency:

- All @threshold tags reference valid PRD thresholds - Threshold values match PRD source - No orphan threshold references - No hardcoded numeric values in steps

1. Cumulative Tag Validation:

- @brd tags reference valid BRD elements - @prd tags reference valid PRD elements - @ears tags reference valid EARS elements - Tags are Gherkin-native (not in comments)

1. Scenario Coverage:

- All 8 scenario categories represented - Success path scenarios for each EARS event-driven statement - Error scenarios for each EARS unwanted behavior statement - Edge cases for boundary conditions - Data-driven scenarios for parameterized tests

5.2 Fix Cycle

If review score < 90%, invoke doc-bdd-fixer.

/doc-bdd-fixer BDD-NN --revalidate

Fix Categories:

Output: BDD-NN.F_fix_report_v001.md

5.3 Re-Review

After fixes, automatically re-run reviewer.

/doc-bdd-reviewer BDD-NN

Output: BDD-NN.A_audit_report_v002.md

5.4 Iteration Control

Iteration Example:

Iteration 1:
  Review v001: Score 84 (3 errors, 6 warnings)
  Fix v001: Fixed 7 issues, added 5 threshold references

Iteration 2:
  Review v002: Score 92 (0 errors, 4 warnings)
  Status: PASS (score >= 90)

5.5 Quality Checks (Post-Fix)

After passing the fix cycle:

1. ADR-Ready Report (with Visual Indicators): Score Display Format: ADR-Ready Score Breakdown ========================= Scenario Completeness: 35/35 ✅ EARS Translation: 15/15 Success/Error/Edge: 15/15 Observable Verification: 5/5 Testability: 30/30 ✅ Automatable Scenarios: 15/15 Data-Driven Examples: 10/10 Performance Benchmarks: 5/5 Architecture Requirements: 25/25 ✅ Quality Attributes: 15/15 Integration Points: 10/10 Business Validation: 10/10 ✅ Acceptance Criteria: 5/5 Success Outcomes: 5/5 v2.0 Compliance: (Bonus) Scenario Type Tags: ✅ All scenarios tagged Priority Tags: ✅ All scenarios prioritized SHALL+WITHIN Language: ✅ 15/15 timed scenarios 5-Category Coverage: ✅ All 5 categories present Threshold Format: ✅ @threshold:PRD.NN format ---------------------------- Total ADR-Ready Score: ✅ 100/100 (Target: >= 90) Status: READY FOR ADR GENERATION

- ✅ >= 90% (Target Met) - 🟡 85-89% (Near Threshold) - ❌ < 85% (Failing)

1. Traceability Matrix Update: # Update BDD-00_TRACEABILITY_MATRIX.md python ai_dev_ssd_flow/scripts/update_traceability_matrix.py \ --bdd docs/04_BDD/BDD-NN_{slug}/ \ --matrix docs/04_BDD/BDD-00_TRACEABILITY_MATRIX.md

Scenario Category Reference

1. Success Path Scenarios (@primary)

Purpose: Validate happy path functionality

Source: EARS Event-Driven statements (WHEN...THE...SHALL)

Example:

@primary @functional
Scenario: User successfully authenticates with valid credentials
  Given user has valid credentials
  And authentication service is available
  When user submits authentication request
  Then user session is created
  And session token is returned
  And response time is less than @threshold:PRD.01.auth.response.p95

Coverage Target: Every EARS event-driven requirement

2. Alternative Path Scenarios (@alternative)

Purpose: Validate optional parameters and alternate workflows

Source: EARS variations and optional paths

Example:

@alternative @optional_params
Scenario: User authenticates with remember-me option
  Given user has valid credentials
  When user submits authentication with remember-me enabled
  Then extended session token is returned
  And session expiry is set to @threshold:PRD.01.session.extended_ttl

Coverage Target: Each optional parameter combination

3. Error Condition Scenarios (@negative)

Purpose: Validate error handling and invalid inputs

Source: EARS Unwanted Behavior statements (IF...THEN...SHALL)

Example:

@negative @error_handling
Scenario: Authentication fails with invalid credentials
  Given user has invalid credentials
  When user submits authentication request
  Then authentication is rejected
  And error code "AUTH_INVALID_CREDENTIALS" is returned
  And failed attempt is logged

Coverage Target: Every EARS unwanted behavior statement

4. Edge Case Scenarios (@edge_case, @boundary)

Purpose: Validate boundary conditions and limits

Source: EARS constraints and threshold boundaries

Example:

@edge_case @boundary
Scenario: Session expires at exact timeout boundary
  Given session was created at "09:00:00" in "America/New_York"
  And session timeout is @threshold:PRD.01.session.idle_timeout
  When current time reaches session expiry
  Then session is invalidated
  And user is prompted to re-authenticate

Coverage Target: Each numeric threshold and boundary condition

5. Data-Driven Scenarios (@data_driven)

Purpose: Validate behavior across multiple data combinations

Source: EARS statements with parameterized values

Example:

@data_driven
Scenario Outline: Validate user role authorization
  Given user has role <role>
  When user attempts to access <resource>
  Then access is <result>

  Examples:
    | role    | resource    | result  |
    | admin   | /admin      | granted |
    | user    | /admin      | denied  |
    | user    | /dashboard  | granted |
    | guest   | /dashboard  | denied  |

Coverage Target: Each parameterized requirement

6. Integration Scenarios (@integration)

Purpose: Validate external system interactions

Source: EARS statements involving external dependencies

Example:

@integration @external_api
Scenario: System integrates with external identity provider
  Given external identity provider is configured
  When user initiates SSO authentication
  Then system redirects to identity provider
  And callback is processed within @threshold:PRD.01.sso.callback_timeout

Coverage Target: Each external system dependency

7. Quality Attribute Scenarios (@quality_attribute)

Purpose: Validate non-functional requirements

Source: EARS Quality Attributes section

Example:

@quality_attribute @performance
Scenario: Authentication endpoint meets performance SLA
  Given system is under normal load
  When 100 concurrent authentication requests are submitted
  Then p95 response time is less than @threshold:PRD.01.perf.auth.p95_latency
  And no requests fail due to timeout

Coverage Target: Each quality attribute (performance, security, reliability)

8. Failure Recovery Scenarios (@failure_recovery)

Purpose: Validate resilience and recovery mechanisms

Source: EARS Unwanted Behavior recovery statements

Example:

@failure_recovery @circuit_breaker
Scenario: Authentication service recovers from database failure
  Given database connection fails
  When authentication request is received
  Then circuit breaker activates
  And fallback authentication is attempted
  And service recovers within @threshold:PRD.01.recovery.max_time

Coverage Target: Each recovery and resilience mechanism

Enhanced Scenario Tagging (v2.0)

Scenario Type Classification

All scenarios MUST include a @scenario-type tag for classification:

Priority Classification

All scenarios MUST include a priority tag:

SHALL+WITHIN Language Pattern

For timed operations, use formal EARS-derived language with WITHIN constraints:

@scenario-type:success @p0-critical
Scenario: Authentication completes within performance threshold
  Given user has valid credentials
  When user submits authentication request
  Then the system SHALL authenticate the user
  And the response SHALL be returned WITHIN @threshold:PRD.01.perf.auth.p95_latency

WITHIN Constraint Rules:

1. All performance-related scenarios MUST include WITHIN clauses
2. WITHIN values MUST reference threshold registry (no hardcoded values)
3. Format: WITHIN @threshold:PRD.NN.category.field

Enhanced Scenario Template

@section: NN.SS
@parent_doc: BDD-NN
@index: BDD-NN.0_index.md
@brd:BRD.NN.01.SS
@prd:PRD.NN.07.SS
@ears:EARS.NN.25.SS

Feature: BDD-NN.SS: [Feature Name]
  As a [role]
  I want [feature]
  So that [benefit]

  Background:
    Given the system timezone is "America/New_York"
    And the current time is "09:30:00" in "America/New_York"

  @scenario-type:success @p0-critical @scenario-id:BDD.NN.14.01
  Scenario: [Primary success path description]
    Given [precondition from EARS WHEN clause]
    When [action from EARS trigger]
    Then the system SHALL [outcome from EARS SHALL clause]
    And the response SHALL be returned WITHIN @threshold:PRD.NN.perf.api.p95_latency

  @scenario-type:error @p1-high @scenario-id:BDD.NN.14.10
  Scenario: [Error condition] results in [expected behavior]
    Given [error precondition from EARS IF clause]
    When [action that triggers error]
    Then the system SHALL NOT [prevented behavior]
    And error code "[ERROR_CODE]" SHALL be returned WITHIN @threshold:PRD.NN.timeout.error.response

  @scenario-type:recovery @p1-high @scenario-id:BDD.NN.14.20
  Scenario: System recovers from [failure type]
    Given [failure condition]
    When [recovery trigger]
    Then the system SHALL recover WITHIN @threshold:PRD.NN.recovery.max_time
    And circuit breaker state SHALL transition to "half-open"

  @scenario-type:parameterized @p2-medium @scenario-id:BDD.NN.14.30
  Scenario Outline: [Parameterized test description]
    Given [context with <variable>]
    When [action with <variable>]
    Then the system SHALL [outcome with <expected>]

    Examples:
      | variable | expected |
      | value1   | result1  |
      | value2   | result2  |

  @scenario-type:optional @p2-medium @scenario-id:BDD.NN.14.40
  Scenario: [Alternative path with optional parameter]
    Given [optional context]
    When [alternative action]
    Then the system SHALL [alternative outcome]

5-Category Coverage Matrix

Execution Modes

Single EARS Mode

Generate BDD from one EARS document.

# Example: Generate BDD from EARS-01
python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
  --ears docs/03_EARS/EARS-01_f1_iam/ \
  --output docs/04_BDD/ \
  --id 01 \
  --slug f1_iam_scenarios

Batch Mode

Generate BDD from multiple EARS in sequence.

# Example: Generate BDD from all EARS
python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
  --batch config/bdd_batch.yaml \
  --output docs/04_BDD/

Batch Configuration (config/bdd_batch.yaml):

bdd_generation:
  - id: "01"
    slug: "f1_iam_scenarios"
    ears: "EARS-01"
    priority: 1

  - id: "02"
    slug: "f2_data_scenarios"
    ears: "EARS-02"
    priority: 2

  - id: "03"
    slug: "f3_api_scenarios"
    ears: "EARS-03"
    priority: 2

execution:
  parallel: false
  fail_fast: true

Dry Run Mode

Preview execution plan without generating files.

python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
  --ears docs/03_EARS/EARS-01_f1_iam/ \
  --dry-run

Review Mode (v2.1)

Validate existing BDD documents and generate a quality report without modification.

Purpose: Audit existing BDD documents for compliance, quality scores, and identify issues.

Command:

# Review single BDD suite
python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
  --bdd docs/04_BDD/BDD-01_f1_iam/ \
  --mode review

# Review all BDD suites
python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
  --bdd docs/04_BDD/ \
  --mode review \
  --output-report tmp/bdd_review_report.md

Review Process:

flowchart TD
    A[Input: Existing BDD] --> B[Load BDD Documents]
    B --> C[Run Validation Checks]
    C --> D[Calculate ADR-Ready Score]
    D --> E[Identify Issues]
    E --> F[Categorize Issues]
    F --> G{Generate Report}
    G --> H[Fixable Issues List]
    G --> I[Manual Review Items]
    G --> J[Score Breakdown]
    H --> K[Output: Review Report]
    I --> K
    J --> K

Review Report Structure:

# BDD Review Report: BDD-01_f1_iam

## Summary
- **ADR-Ready Score**: 87% 🟡
- **Total Issues**: 12
- **Auto-Fixable**: 8
- **Manual Review**: 4

## Score Breakdown
| Category | Score | Status |
|----------|-------|--------|
| Scenario Completeness | 32/35 | 🟡 |
| Testability | 28/30 | ✅ |
| Architecture Requirements | 20/25 | 🟡 |
| Business Validation | 7/10 | ❌ |

## v2.0 Compliance
| Check | Status | Details |
|-------|--------|---------|
| Scenario Type Tags | ❌ | 5 scenarios missing @scenario-type |
| Priority Tags | 🟡 | 3 scenarios missing priority |
| SHALL+WITHIN Language | ❌ | 8 timed scenarios lack WITHIN |
| 5-Category Coverage | ✅ | All categories present |
| Threshold Format | 🟡 | 2 hardcoded values found |

## Auto-Fixable Issues
| # | Issue | Location | Fix Action |
|---|-------|----------|------------|
| 1 | Missing @scenario-type tag | BDD-01.1:L45 | Add @scenario-type:success |
| 2 | Hardcoded timeout value | BDD-01.2:L78 | Replace with @threshold |
| ... | ... | ... | ... |

## Manual Review Required
| # | Issue | Location | Reason |
|---|-------|----------|--------|
| 1 | Ambiguous step definition | BDD-01.3:L102 | Requires domain knowledge |
| 2 | Missing edge case scenario | BDD-01.1 | Content decision needed |
| ... | ... | ... | ... |

## Recommendations
1. Run fix mode to address 8 auto-fixable issues
2. Review and update 4 items requiring manual attention
3. Add 2 missing edge case scenarios for full coverage

Review Configuration:

review_mode:
  enabled: true
  checks:
    - structure_validation      # Index, sections, redirects
    - gherkin_syntax           # Given-When-Then compliance
    - tag_validation           # Cumulative tags, scenario tags
    - threshold_references     # @threshold format and validity
    - v2_compliance            # Scenario types, priorities, WITHIN
    - coverage_analysis        # 5-category coverage
    - score_calculation        # ADR-Ready score
  output:
    format: markdown           # markdown, json, html
    include_line_numbers: true
    include_fix_suggestions: true
  thresholds:
    pass: 90
    warning: 85
    fail: 0

Fix Mode (v2.1)

Auto-repair existing BDD documents while preserving manual content.

Purpose: Apply automated fixes to BDD documents to improve quality scores and compliance.

Command:

# Fix single BDD suite
python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
  --bdd docs/04_BDD/BDD-01_f1_iam/ \
  --mode fix

# Fix with backup
python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
  --bdd docs/04_BDD/BDD-01_f1_iam/ \
  --mode fix \
  --backup

# Fix specific issue types only
python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
  --bdd docs/04_BDD/BDD-01_f1_iam/ \
  --mode fix \
  --fix-types "tags,thresholds,syntax"

# Dry-run fix (preview changes)
python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
  --bdd docs/04_BDD/BDD-01_f1_iam/ \
  --mode fix \
  --dry-run

Fix Process:

flowchart TD
    A[Input: Existing BDD] --> B[Run Review Mode]
    B --> C[Identify Fixable Issues]
    C --> D{Backup Enabled?}
    D -->|Yes| E[Create Backup]
    D -->|No| F[Skip Backup]
    E --> G[Apply Fixes by Category]
    F --> G

    subgraph FixCategories["Fix Categories"]
        G --> H[Structure Fixes]
        G --> I[Tag Fixes]
        G --> J[Threshold Fixes]
        G --> K[Syntax Fixes]
        G --> L[v2.0 Compliance Fixes]
    end

    H --> M[Preserve Manual Content]
    I --> M
    J --> M
    K --> M
    L --> M

    M --> N[Re-validate]
    N --> O{Score Improved?}
    O -->|Yes| P[Generate Fix Report]
    O -->|No| Q[Log Warnings]
    Q --> P
    P --> R[Output: Fixed BDD + Report]

Fix Categories and Actions:

Content Preservation Rules:

1. Never delete existing scenario content
2. Never modify step definitions (Given/When/Then text)
3. Never change Examples table data
4. Only add missing tags and metadata
5. Only replace hardcoded values with threshold references
6. Backup first if --backup flag is set

Fix Report Structure:

# BDD Fix Report: BDD-01_f1_iam

## Summary
- **Before ADR-Ready Score**: 87% 🟡
- **After ADR-Ready Score**: 94% ✅
- **Issues Fixed**: 8
- **Issues Remaining**: 4 (manual review required)

## Fixes Applied
| # | Issue | Location | Fix Applied |
|---|-------|----------|-------------|
| 1 | Missing @scenario-type | BDD-01.1:L45 | Added @scenario-type:success |
| 2 | Hardcoded timeout | BDD-01.2:L78 | Replaced with @threshold:PRD.01.timeout.api.max |
| 3 | Tags in comments | BDD-01.3:L12 | Moved to Gherkin-native position |
| ... | ... | ... | ... |

## Files Modified
- docs/04_BDD/BDD-01_f1_iam/BDD-01.1_authentication.feature
- docs/04_BDD/BDD-01_f1_iam/BDD-01.2_session.feature
- docs/04_BDD/BDD-01_f1_iam/BDD-01.3_authorization.feature

## Backup Location
- tmp/backup/BDD-01_f1_iam_20260209_143022/

## Remaining Issues (Manual Review)
| # | Issue | Location | Reason |
|---|-------|----------|--------|
| 1 | Ambiguous step | BDD-01.3:L102 | Requires domain knowledge |
| ... | ... | ... | ... |

## Next Steps
1. Review manually flagged items
2. Re-run validation to confirm score
3. Commit changes if satisfied

Fix Configuration:

fix_mode:
  enabled: true
  backup:
    enabled: true
    location: "tmp/backup/"
    retention_days: 7

  fix_categories:
    structure: true          # Index, redirects
    tags: true               # Scenario tags, cumulative tags
    thresholds: true         # @threshold references
    syntax: false            # Gherkin syntax (risky, disabled by default)
    v2_compliance: true      # Scenario types, priorities, WITHIN

  preservation:
    step_definitions: true   # Never modify step text
    examples_data: true      # Never modify Examples tables
    comments: true           # Preserve user comments
    custom_tags: true        # Preserve non-standard tags

  validation:
    re_validate_after_fix: true
    require_score_improvement: false
    max_fix_iterations: 3

Command Line Options (Review/Fix):

Configuration

Default Configuration

# config/bdd_autopilot.yaml
bdd_autopilot:
  version: "2.0"

  scoring:
    bdd_ready_min: 90
    adr_ready_min: 90
    strict_mode: false
    # NEW: Visual score indicators
    score_display:
      pass_icon: "✅"      # >= 90%
      warning_icon: "🟡"   # 85-89%
      fail_icon: "❌"      # < 85%

  execution:
    max_parallel: 3        # HARD LIMIT - do not exceed
    chunk_size: 3          # Documents per chunk
    pause_between_chunks: true
    auto_fix: true
    continue_on_error: false
    timeout_per_ears: 300  # seconds

  output:
    structure: sectioned  # always sectioned for BDD
    size_threshold_lines: 800
    max_scenarios_per_feature: 12
    report_format: markdown

  validation:
    skip_validation: false
    fix_iterations_max: 3
    # NEW: Enhanced validation checks
    require_scenario_type_tags: true
    require_priority_tags: true
    require_within_constraints: true
    require_5_category_coverage: true

  review:
    enabled: true
    check_gherkin: true
    check_thresholds: true
    check_coverage: true
    check_tags: true
    auto_fix_gherkin: true
    # NEW: SHALL+WITHIN enforcement
    check_timing_constraints: true

  coverage:
    require_success_paths: true
    require_error_scenarios: true
    require_edge_cases: true
    require_data_driven: true
    require_quality_attributes: true
    # NEW: 5-Category Coverage
    categories:
      - success      # Primary happy path
      - optional     # Alternative paths
      - recovery     # Failure recovery
      - parameterized # Data-driven
      - error        # Negative cases

  # NEW: Scenario Classification
  scenario_classification:
    type_tags:
      - "@scenario-type:success"
      - "@scenario-type:optional"
      - "@scenario-type:recovery"
      - "@scenario-type:parameterized"
      - "@scenario-type:error"
    priority_tags:
      - "@p0-critical"  # MVP blocking
      - "@p1-high"      # Sprint required
      - "@p2-medium"    # Next iteration
      - "@p3-low"       # Backlog

  # NEW: Threshold Reference Format
  threshold_format:
    pattern: "@threshold:PRD.NN.category.field"
    examples:
      - "@threshold:PRD.01.perf.auth.p95_latency"
      - "@threshold:PRD.01.timeout.session.idle"
      - "@threshold:PRD.01.retry.max_attempts"

Command Line Options

Context Management

Chunked Parallel Execution (MANDATORY)

CRITICAL: To prevent conversation context overflow errors ("Prompt is too long", "Conversation too long"), all autopilot operations MUST follow chunked execution rules:

Chunk Size Limit: Maximum 3 documents per chunk

Chunking Rules:

1. Chunk Formation: Group EARS documents into chunks of maximum 3 at a time
2. Sequential Chunk Processing: Process one chunk at a time, completing all BDDs in a chunk before starting the next
3. Context Pause: After completing each chunk, provide a summary and pause for user acknowledgment
4. Progress Tracking: Display chunk progress (e.g., "Chunk 2/5: Processing EARS-04, EARS-05, EARS-06")

Execution Pattern:

EARS: EARS-01, EARS-02, EARS-03, EARS-04, EARS-05, EARS-06, EARS-07

Chunk 1: [EARS-01, EARS-02, EARS-03] → Process → Summary → Pause
Chunk 2: [EARS-04, EARS-05, EARS-06] → Process → Summary → Pause
Chunk 3: [EARS-07]                    → Process → Summary → Complete

Chunk Completion Summary Template:

Chunk N/M Complete:
├── Generated: BDD-01, BDD-02, BDD-03
├── ADR-Ready Scores: 94%, 92%, 91%
├── Status: All validated
└── Next: Continue with Chunk N+1? [Y/n]

Why Chunking is Required:

• Prevents "Conversation too long" errors during batch processing
• Allows context compaction between chunks
• Enables recovery from failures without losing all progress
• Provides natural checkpoints for user review

Output Artifacts

Generated Files

All BDD suites use nested folders (BDD-NN_{slug}/). This keeps feature files and companion files (review reports, fix reports, drift cache) organized together.

Section File Structure

Validation Reports

Error Handling

Error Categories

Recovery Actions

def handle_error(error_type: str, context: dict) -> Action:
    match error_type:
        case "EARS_MISSING":
            return Action.ABORT_WITH_MESSAGE
        case "BDD_READY_LOW":
            return Action.AUTO_FIX_EARS
        case "VALIDATION_FAILURE":
            if context["retry_count"] < 3:
                return Action.AUTO_FIX_RETRY
            return Action.FLAG_MANUAL_REVIEW
        case "ADR_READY_LOW":
            return Action.ENHANCE_SCENARIOS
        case _:
            return Action.FLAG_MANUAL_REVIEW

Integration Points

Pre-Execution Hooks

# Hook: pre_bdd_generation
# Runs before BDD generation starts
./hooks/pre_bdd_generation.sh

# Example: Validate EARS exists and is ready
if [ ! -d "docs/03_EARS/EARS-01_*" ]; then
  echo "ERROR: EARS-01 required"
  exit 1
fi

Post-Execution Hooks

# Hook: post_bdd_generation
# Runs after BDD generation completes
./hooks/post_bdd_generation.sh

# Example: Trigger ADR autopilot for validated BDD
if [ "$BDD_VALIDATED" = "true" ]; then
  python ai_dev_ssd_flow/05_ADR/scripts/adr_autopilot.py \
    --bdd "$BDD_PATH" \
    --output docs/05_ADR/
fi

CI/CD Integration

# .github/workflows/bdd_autopilot.yml
name: BDD Autopilot

on:
  push:
    paths:
      - 'docs/03_EARS/**'

jobs:
  generate-bdd:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4

      - name: Run BDD Autopilot
        run: |
          python ai_dev_ssd_flow/04_BDD/scripts/bdd_autopilot.py \
            --ears docs/03_EARS/ \
            --output docs/04_BDD/ \
            --validate

      - name: Upload Validation Report
        uses: actions/upload-artifact@v4
        with:
          name: bdd-validation
          path: tmp/bdd_validation_report.json

Quality Gates

Phase Gates

Blocking vs Non-Blocking

Validation Checklist

After autopilot completion:

• All target BDD suites generated
• Each BDD has ADR-Ready score >= 90%
• Traceability matrix updated (BDD-00_TRACEABILITY_MATRIX.md)
• Each BDD references upstream EARS with @ears tags
• Each BDD references upstream PRD with @prd tags
• Each BDD references upstream BRD with @brd tags
• All scenarios use Given-When-Then structure
• All quantitative values use @threshold references
• Tags are Gherkin-native (not in comments)
• Times include seconds (HH:MM:SS) with IANA timezone
• All 8 scenario categories represented
• No.feature file exceeds 800 lines
• No Feature block exceeds 12 scenarios
• Element IDs use BDD.NN.14.SS (scenarios), BDD.NN.15.SS (steps) format
• Index file (BDD-NN.0_index.md) exists with section file map
• Redirect stub exists at docs/04_BDD/ root

Quick Reference

Usage:

/doc-bdd-autopilot <EARS-LIST> [OPTIONS]

Common Commands:

# Single EARS
/doc-bdd-autopilot EARS-01

# All EARS (automatic)
/doc-bdd-autopilot all --auto

# Preview only
/doc-bdd-autopilot all --dry-run

# Resume after failure
/doc-bdd-autopilot resume

Related Resources

Skills (Delegated)

• BDD Skill: .claude/skills/doc-bdd/SKILL.md - BDD creation rules and Gherkin syntax
• BDD Validator Skill: .claude/skills/doc-bdd-validator/SKILL.md - Validation rules and error codes
• EARS Validator Skill: .claude/skills/doc-ears-validator/SKILL.md - EARS BDD-Ready validation
• Quality Advisor Skill: .claude/skills/quality-advisor/SKILL.md - Real-time quality feedback
• Naming Standards Skill: .claude/skills/doc-naming/SKILL.md - Element ID format

Templates and Rules

• BDD Template: ai_dev_ssd_flow/04_BDD/BDD-MVP-TEMPLATE.feature
• Index Template: ai_dev_ssd_flow/04_BDD/BDD-SECTION-0-TEMPLATE.md
• Aggregator Template: ai_dev_ssd_flow/04_BDD/BDD-AGGREGATOR-TEMPLATE.feature
• BDD Schema: ai_dev_ssd_flow/04_BDD/BDD_MVP_SCHEMA.yaml
• BDD Creation Rules: ai_dev_ssd_flow/04_BDD/BDD-MVP-TEMPLATE.feature
• BDD Validation Rules: ai_dev_ssd_flow/04_BDD/BDD_MVP_SCHEMA.yaml

Framework References

• SDD Workflow: ai_dev_ssd_flow/AI_ASSISTANT_PLAYBOOK.md
• MVP Autopilot: ai_dev_ssd_flow/MVP_AUTOPILOT.md
• EARS Autopilot Skill: .claude/skills/doc-ears-autopilot/SKILL.md
• ADR Autopilot Skill: .claude/skills/doc-adr-autopilot/SKILL.md

Review Document Standards (v2.2)

IMPORTANT: Review reports generated by this autopilot are formal project documents.

See: .claude/skills/REVIEW_DOCUMENT_STANDARDS.md for complete standards.

Quick Reference

Example Location:

docs/04_BDD/
├── BDD-03_f3_observability.feature
├── BDD-03.A_audit_report_v001.md    # ← Preferred report stored here
└── BDD-03.R_review_report_v001.md    # ← Legacy-compatible reviewer report

Version History