Doc Spec Autopilot

doc-spec-autopilot

Purpose

Automated Technical Specifications (SPEC) generation pipeline that processes REQ documents (and optional CTR) to generate implementation-ready YAML specifications with TASKS-Ready scoring.

Layer: 9

Upstream: REQ (Layer 7), CTR (Layer 8 - optional)

Downstream: TSPEC (Layer 10), TASKS (Layer 11)

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

Smart Document Detection

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

Input Type Recognition (Multiple Upstreams)

SPEC can be derived from REQ and/or CTR:

Detection Algorithm

1. Parse input: Extract TYPE and NN from "{TYPE}-{NN}"
2. Determine action:
   - IF TYPE == "SPEC": Review Mode
   - ELSE IF TYPE in ["REQ", "CTR"]: Generate/Find Mode
   - ELSE: Error (invalid type for this autopilot)
3. For Generate/Find Mode:
   - Check: Does SPEC-{NN} exist in docs/09_SPEC/?
   - IF exists: Switch to Review Mode for SPEC-{NN}
   - ELSE: Proceed with Generation from {TYPE}-{NN}

File Existence Check

# Check for nested folder structure (mandatory)
ls docs/09_SPEC/SPEC-{NN}_*/

Examples

# Review mode (same type - SPEC input)
/doc-spec-autopilot SPEC-01          # Reviews existing SPEC-01

# Generate/Find mode (upstream types)
/doc-spec-autopilot REQ-01           # Generates SPEC-01 if missing, or reviews existing SPEC-01
/doc-spec-autopilot CTR-01           # Generates SPEC-01 if missing, or reviews existing SPEC-01

# Multiple inputs
/doc-spec-autopilot REQ-01,REQ-02    # Generates/reviews SPEC-01 and SPEC-02
/doc-spec-autopilot SPEC-01,SPEC-02  # Reviews SPEC-01 and SPEC-02

Action Determination Output

Input: REQ-01
├── Detected Type: REQ (primary upstream)
├── Expected SPEC: SPEC-01
├── SPEC Exists: Yes → docs/09_SPEC/SPEC-01_f1_iam/
└── Action: REVIEW MODE - Running doc-spec-reviewer on SPEC-01

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

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

Workflow Overview

flowchart TD
    subgraph Phase1["Phase 1: REQ/CTR Analysis"]
        A[Start] --> B[Read REQ Documents]
        B --> C[Read CTR Documents if exists]
        C --> D[Extract Implementation Details]
        D --> E[Map to SPEC Structure]
    end

    subgraph Phase2["Phase 2: SPEC Readiness Check"]
        E --> F[Check REQ SPEC-Ready Score]
        F --> G{Score >= 90%?}
        G -->|No| H[Flag REQ Issues]
        H --> I{Auto-Fixable?}
        I -->|Yes| J[Fix REQ Issues]
        J --> F
        I -->|No| K[Abort - Manual Fix Required]
        G -->|Yes| L[Mark REQ Ready]
    end

    subgraph Phase3["Phase 3: SPEC Generation"]
        L --> M[Generate YAML Structure]
        M --> N[Add Methods/Interfaces]
        N --> O[Add Data Models]
        O --> P[quality-advisor: Real-time Feedback]
        P --> Q[Add Traceability]
        Q --> R[Write SPEC File]
    end

    subgraph Phase4["Phase 4: SPEC Validation"]
        R --> S[Run doc-spec-validator]
        S --> T{TASKS-Ready >= 90%?}
        T -->|No| U[Auto-Fix SPEC Issues]
        U --> V[Re-validate SPEC]
        V --> T
        T -->|Yes| W[Mark SPEC Validated]
    end

    subgraph Phase5["Phase 5: Review & Fix Cycle"]
        W --> X[Run doc-spec-reviewer]
        X --> X2{Score >= 90?}
        X2 -->|No| X3[Run doc-spec-fixer]
        X3 --> X4{Iteration < Max?}
        X4 -->|Yes| X
        X4 -->|No| X5[Flag Manual Review]
        X2 -->|Yes| Y[Verify Quality Checks]
        X5 --> Y
        Y --> Z[Update Traceability Matrix]
        Z --> AA[Generate Summary Report]
    end

    AA --> AB[Complete]
    K --> AC[Exit with Error]

SPEC YAML Format (13 Required Sections)

Trading Nexus patterns require comprehensive 13-section YAML structure:

# SPEC-NN: Specification Title
# Required sections (all 13 mandatory)

metadata:
  spec_id: SPEC-01
  title: "Component Specification"
  version: "1.0.0"
  status: "approved"
  created_date: "2026-02-09T00:00:00"
  last_updated: "2026-02-09T00:00:00"
  task_ready_score: "✅ 95% (Target: ≥90%)"
  authors: [{name: "...", email: "...", role: "..."}]
  reviewers: [{name: "...", email: "...", role: "..."}]

traceability:
  upstream_sources:
    business_requirements:
      - id: "BRD.01.01.03"
        link: "../01_BRD/BRD-01.md#BRD.01.01.03"
        relationship: "Business driver"
    product_requirements:
      - id: "PRD.01.07.02"
        link: "../02_PRD/PRD-01.md#PRD.01.07.02"
    atomic_requirements:
      - id: "REQ-01.01.01"
        # CRITICAL: Use nested REQ path format
        link: "../07_REQ/SYS-01_iam/REQ-01.01_jwt_authentication.md"
  cumulative_tags:
    brd: ["BRD.01.01.03"]
    prd: ["PRD.01.07.02"]
    ears: ["EARS.01.25.01"]
    bdd: ["BDD.01.14.01"]
    adr: ["ADR-01"]
    sys: ["SYS.01.26.01"]
    req: ["REQ.01.27.01"]
    ctr: ["CTR.01.16.01"]
    threshold: ["perf.auth.p95_latency", "sla.uptime.target"]  # 9th layer

interfaces:
  # Level 1: External APIs (REST)
  external_apis:
    - endpoint: "POST /api/v1/auth/login"
      method: "POST"
      auth: "None"
      rate_limit: "5 req/5 min per IP"
      request_schema:
        type: "object"
        required: ["email", "password"]
        properties:
          email: { type: "string", format: "email" }
          password: { type: "string", minLength: 12 }
      response_schema:
        type: "object"
        properties:
          access_token: { type: "string" }
          token_type: { type: "string", enum: ["Bearer"] }
      latency_target_ms: "@threshold:perf.auth.p95_latency"

  # Level 2: Internal APIs (Service signatures)
  internal_apis:
    - interface: "AuthService.authenticate()"
      signature: "async def authenticate(email: str, password: str) -> TokenPair"
      purpose: |
        1. Fetch user by email from Identity Platform.
        2. Verify password via Argon2id.
        3. Issue JWT token pair.

  # Level 3: Classes (OOP structure)
  classes:
    - name: "IAMService"
      description: "Facade combining auth, token, and authz services"
      constructor:
        params:
          config: { type: object, required: true }
      methods:
        - name: "initialize"
          input: { }
          output: { success: boolean }

data_models:
  - id: SPEC.01.17.01
    name: "RequestModel"
    json_schema:
      type: object
      properties:
        id: { type: string }
    pydantic_code: |
      from pydantic import BaseModel
      class RequestModel(BaseModel):
          id: str

validation_rules:
  - id: SPEC.01.21.01
    rule: "Email format validation"
    implementation: "Use EmailStr from pydantic"

error_handling:
  catalog:
    INVALID_CREDENTIALS:
      http_status: 401
      message: "Invalid email or password"
    RATE_LIMIT_EXCEEDED:
      http_status: 429
      message: "Too many attempts"

configuration:
  environment_variables:
    - name: JWT_SECRET
      required: true
      description: "JWT signing secret"
  feature_flags:
    - name: MFA_REQUIRED
      default: false

performance:
  latency_targets:
    login_p95_ms: "@threshold:perf.auth.p95_latency"
  throughput_targets:
    auth_rps: "@threshold:perf.auth.throughput"

behavior:
  login_flow:
    pseudocode: |
      def login(email, password):
          enforce_rate_limit(ip, "login")
          user = idp.get_user(email)
          if not user or not verify_password(password, user.hash):
              raise AuthenticationError("INVALID_CREDENTIALS")
          return token_service.issue_tokens(user)

behavioral_examples:
  api_example_login:
    request:
      endpoint: "/api/v1/auth/login"
      payload: { email: "user@example.com", password: "SecureP@ss!" }
    response:
      status: 200
      body: { access_token: "<jwt>", token_type: "Bearer" }

architecture:
  component_structure:
    - name: "AuthService"
      responsibility: "User authentication"
      dependencies: ["GCP Identity Platform", "Redis"]
  resilience:
    circuit_breaker_enabled: true
    retry_policy:
      max_attempts: 3
      backoff_strategy: "exponential"

operations:
  slo:
    uptime: "@threshold:sla.uptime.target"
    error_rate: "<1%"
  monitoring_metrics: ["auth_login_latency_p95_ms", "auth_errors_total"]

req_implementations:
  - req_id: "REQ-01.01"
    req_link: "../07_REQ/SYS-01_iam/REQ-01.01.md"
    implementation:
      interfaces:
        - class: "AuthService"
          method: "login"
          signature: "async def login(...) -> LoginResult"
      data_models:
        - name: "LoginRequest"
          fields: ["email", "password"]
      validation_rules:
        - rule: "Rate limit login attempts"
          implementation: "5 attempts/5 minutes per IP"
      error_handling:
        - error_code: "INVALID_CREDENTIALS"
          http_status: 401
      test_approach:
        unit_tests: ["hash verification rejects wrong password"]
        integration_tests: ["login flow returns token pair"]

threshold_references:
  registry_document: "PRD-01_thresholds"
  keys_used:
    performance:
      - key: "perf.auth.p95_latency"
        usage: "performance.latency_targets.login"
    sla:
      - key: "sla.uptime.target"
        usage: "operations.slo.uptime"

Phase 5: Review & Fix Cycle (v2.2)

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

flowchart TD
    A[Phase 5 Start] --> B[Run doc-spec-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-spec-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-spec-reviewer to identify issues.

/doc-spec-reviewer SPEC-NN

Output: SPEC-NN.R_review_report_v001.md

5.2 Fix Cycle

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

/doc-spec-fixer SPEC-NN --revalidate

Fix Categories:

Output: SPEC-NN.F_fix_report_v001.md

5.3 Re-Review

After fixes, automatically re-run reviewer.

/doc-spec-reviewer SPEC-NN

Output: SPEC-NN.R_review_report_v002.md

5.4 Iteration Control

Iteration Example:

Iteration 1:
  Review v001: Score 82 (3 errors, 5 warnings)
  Fix v001: Fixed 6 issues, added 2 sections

Iteration 2:
  Review v002: Score 93 (0 errors, 3 warnings)
  Status: PASS (score >= 90)

5.5 Quality Checks (Post-Fix)

After passing the fix cycle:

1. Section Completeness:

   • All 13 required sections present
   • No placeholder text remaining ([TBD], TODO, XXX)
   • Minimum content length per section

2. Three-Level Interface Coverage:

   • external_apis defined with OpenAPI format
   • internal_apis defined with method signatures
   • classes defined with constructors and methods

3. Element ID Compliance (per doc-naming skill):

   • All IDs use SPEC.NN.TT.SS format
   • Element type codes valid for SPEC (15, 16, 17, 21, 28)
   • No legacy patterns

4. TASKS-Ready Report:

   TASKS-Ready Score Breakdown
   ===========================
   Interface Completeness:  23/25 (3 levels defined)
   Data Models:             20/20 (Pydantic + JSON Schema)
   Validation Rules:        15/15 (input/output validated)
   Error Handling:          15/15 (catalog with HTTP status)
   Test Approach:           10/10 (unit + integration tests)
   Traceability:            10/10 (all 9 cumulative tags)
   Performance Specs:        5/5 (@threshold references)
   ----------------------------
   Total TASKS-Ready Score: 98/100 (Target: >= 90)
   Status: READY FOR TASKS GENERATION
   

5. Traceability Matrix Update:

   # Update SPEC traceability
   python ai_dev_ssd_flow/scripts/update_traceability_matrix.py \
     --spec docs/09_SPEC/SPEC-NN_{slug}/SPEC-NN_{slug}.yaml \
     --matrix docs/09_SPEC/SPEC-00_TRACEABILITY_MATRIX.md
   

Element Type Codes

Cumulative Tags (7-8 Required)

@brd: BRD.NN.TT.SS
@prd: PRD.NN.TT.SS
@ears: EARS.NN.TT.SS
@bdd: BDD.NN.TT.SS
@adr: ADR-NN
@sys: SYS.NN.TT.SS
@req: REQ.NN.TT.SS
@ctr: CTR.NN.TT.SS  # Optional

Configuration

Default Configuration

spec_autopilot:
  version: "2.0"

  scoring:
    spec_ready_min: 90
    tasks_ready_min: 90
    strict_mode: false
    # NEW: 7-component scoring weights
    scoring_weights:
      interface_completeness: 25  # All 3 levels defined
      data_models: 20             # Pydantic + JSON Schema
      validation_rules: 15        # Input/output validation
      error_handling: 15          # Error catalog with HTTP status
      test_approach: 10           # Unit + integration tests
      traceability: 10            # All 9 cumulative tags
      performance_specs: 5        # Latency targets with thresholds

  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_req: 180  # seconds

  output:
    format: yaml
    report_format: markdown
    # NEW: File splitting strategy
    max_file_size_kb: 66
    split_strategy: auto  # auto, single, nested

  validation:
    skip_validation: false
    fix_iterations_max: 3
    # NEW: Enhanced validation
    require_all_13_sections: true
    require_three_interface_levels: true
    require_threshold_registry: true
    require_req_implementations: true
    require_nested_req_paths: true

  traceability:
    # NEW: 9-layer cumulative tags
    required_layers: 9
    include_threshold_references: true

7-Component TASKS-Ready Scoring

Score Display Format:

✅ 95% (Target: ≥90%)  # Passing
🟡 87% (Target: ≥90%)  # Near threshold
❌ 75% (Target: ≥90%)  # Failing

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 REQ-derived SPEC documents into chunks of maximum 3 at a time
2. Sequential Chunk Processing: Process one chunk at a time, completing all documents 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 2/4: Processing SPEC-04, SPEC-05, SPEC-06...")

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

Chunk Completion Template:

## Chunk N/M Complete

Generated:
- SPEC-XX: TASKS-Ready Score 94%
- SPEC-YY: TASKS-Ready Score 92%
- SPEC-ZZ: TASKS-Ready Score 95%

Proceeding to next chunk...

Execution Modes

Single SPEC Mode

Generate SPEC from one REQ document.

/doc-spec-autopilot REQ-01 --output docs/09_SPEC/

Batch Mode

Generate SPEC from multiple REQ documents.

/doc-spec-autopilot all --auto

Dry Run Mode

Preview execution plan without generating files.

/doc-spec-autopilot REQ-01 --dry-run

Review Mode (v2.1)

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

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

Command:

# Review single SPEC
/doc-spec-autopilot SPEC-01 --mode review

# Review all SPECs
/doc-spec-autopilot all --mode review \
  --output-report tmp/spec_review_report.md

Review Process:

flowchart TD
    A[Input: Existing SPEC] --> B[Load YAML Files]
    B --> C[Validate YAML Syntax]
    C --> D[Check 13 Required Sections]
    D --> E[Validate 3-Level Interfaces]
    E --> F[Check Threshold References]
    F --> G[Calculate TASKS-Ready Score]
    G --> H{Generate Report}
    H --> I[Fixable Issues List]
    H --> J[Manual Review Items]
    H --> K[7-Component Breakdown]
    I --> L[Output: Review Report]
    J --> L
    K --> L

Review Report Structure:

# SPEC Review Report: SPEC-01.yaml

## Summary
- **TASKS-Ready Score**: 87% 🟡
- **Total Issues**: 9
- **Auto-Fixable**: 6
- **Manual Review**: 3

## 7-Component Score Breakdown
| Component | Score | Status |
|-----------|-------|--------|
| Interface Completeness | 23/25 | 🟡 |
| Data Models | 18/20 | 🟡 |
| Validation Rules | 14/15 | ✅ |
| Error Handling | 13/15 | 🟡 |
| Test Approach | 9/10 | ✅ |
| Traceability | 8/10 | 🟡 |
| Performance Specs | 4/5 | 🟡 |

## Section Completeness
| Section | Present | Status |
|---------|---------|--------|
| metadata | ✅ | Complete |
| traceability | ✅ | Missing @ctr tag |
| interfaces | 🟡 | Missing internal_apis |
| data_models | ✅ | Complete |
| validation_rules | ✅ | Complete |
| error_handling | ✅ | Complete |
| configuration | ✅ | Complete |
| performance | 🟡 | Hardcoded values |
| behavior | ✅ | Complete |
| behavioral_examples | ✅ | Complete |
| architecture | ✅ | Complete |
| operations | ✅ | Complete |
| req_implementations | ❌ | Missing |

## v2.0 Compliance
| Check | Status | Details |
|-------|--------|---------|
| 13 Sections Present | ❌ | Missing req_implementations |
| Three Interface Levels | 🟡 | Missing internal_apis |
| Threshold Registry | ✅ | Present |
| Nested REQ Paths | ✅ | Correct format |
| 9-Layer Traceability | 🟡 | Missing @ctr |
| @threshold Format | ❌ | 2 hardcoded values |

## Auto-Fixable Issues
| # | Issue | Location | Fix Action |
|---|-------|----------|------------|
| 1 | Missing @ctr tag | traceability | Add placeholder @ctr reference |
| 2 | Hardcoded latency | performance.latency_targets | Replace with @threshold:perf.api.p95 |
| 3 | Missing req_implementations | root | Add template section |
| ... | ... | ... | ... |

## Manual Review Required
| # | Issue | Location | Reason |
|---|-------|----------|--------|
| 1 | Missing internal_apis | interfaces | Architecture decision needed |
| 2 | Incomplete behavior | behavior.login_flow | Domain knowledge required |
| ... | ... | ... | ... |

Review Configuration:

review_mode:
  enabled: true
  checks:
    - yaml_syntax              # Valid YAML structure
    - section_completeness     # All 13 sections present
    - interface_levels         # 3-level interface hierarchy
    - threshold_references     # @threshold format
    - cumulative_tags          # 9-layer traceability
    - score_calculation        # TASKS-Ready score
    - file_size                # <66KB check
  output:
    format: markdown
    include_fix_suggestions: true
  thresholds:
    pass: 90
    warning: 85
    fail: 0

Fix Mode (v2.1)

Auto-repair existing SPEC documents while preserving manual content.

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

Command:

# Fix single SPEC
/doc-spec-autopilot SPEC-01 --mode fix

# Fix with backup
/doc-spec-autopilot SPEC-01 --mode fix \
  --backup

# Fix specific issue types only
/doc-spec-autopilot SPEC-01 --mode fix \
  --fix-types "sections,thresholds,traceability"

# Dry-run fix (preview changes)
/doc-spec-autopilot SPEC-01 --mode fix \
  --dry-run

Fix Categories and Actions:

Content Preservation Rules:

1. Never delete existing interface definitions
2. Never modify method signatures or logic
3. Never change data model schemas
4. Only add missing sections and metadata
5. Only replace hardcoded values with threshold references
6. Backup first if --backup flag is set

Fix Report Structure:

# SPEC Fix Report: SPEC-01.yaml

## Summary
- **Before TASKS-Ready Score**: 87% 🟡
- **After TASKS-Ready Score**: 94% ✅
- **Issues Fixed**: 6
- **Issues Remaining**: 3 (manual review required)

## Fixes Applied
| # | Issue | Location | Fix Applied |
|---|-------|----------|-------------|
| 1 | Missing req_implementations | root | Added template section |
| 2 | Hardcoded latency | performance.latency_targets | Replaced with @threshold:perf.api.p95 |
| 3 | Missing @ctr tag | traceability | Added @ctr: CTR.01.16.01 |
| ... | ... | ... | ... |

## Files Modified
- docs/09_SPEC/SPEC-01_f1_iam/SPEC-01_f1_iam.yaml

## Backup Location
- tmp/backup/SPEC-01_20260209_143022.yaml

## 7-Component Score Impact
| Component | Before | After | Delta |
|-----------|--------|-------|-------|
| Interface Completeness | 23/25 | 24/25 | +1 |
| Data Models | 18/20 | 18/20 | 0 |
| Validation Rules | 14/15 | 14/15 | 0 |
| Error Handling | 13/15 | 13/15 | 0 |
| Test Approach | 9/10 | 10/10 | +1 |
| Traceability | 8/10 | 10/10 | +2 |
| Performance Specs | 4/5 | 5/5 | +1 |

## Next Steps
1. Add internal_apis level to interfaces section
2. Complete behavior.login_flow pseudocode
3. Re-run validation to confirm score

Fix Configuration:

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

  fix_categories:
    sections: true           # Missing required sections
    thresholds: true         # @threshold references
    traceability: true       # Cumulative tags
    interfaces: false        # Interface placeholders (risky)
    yaml: true               # YAML formatting

  preservation:
    interfaces: true         # Never modify existing interfaces
    data_models: true        # Never modify schemas
    behavior: true           # Never modify logic
    comments: true           # Preserve YAML comments

  validation:
    re_validate_after_fix: true
    require_score_improvement: false
    max_fix_iterations: 3

Command Line Options (Review/Fix):

Related Resources

• SPEC Skill: .claude/skills/doc-spec/SKILL.md
• SPEC Validator: .claude/skills/doc-spec-validator/SKILL.md
• Naming Standards: .claude/skills/doc-naming/SKILL.md
• Quality Advisor: .claude/skills/quality-advisor/SKILL.md
• SPEC Template: ai_dev_ssd_flow/09_SPEC/SPEC-MVP-TEMPLATE.yaml

Validation Rules (v2.0)

File Splitting Strategy

Review Document Standards (v2.2)

Review reports generated by this skill are formal project documents and MUST comply with shared standards.

Reference: See REVIEW_DOCUMENT_STANDARDS.md in the skills directory for complete requirements.

Key Requirements:

1. Storage Location: Same folder as the reviewed SPEC document
2. File Naming: preferred SPEC-NN.A_audit_report_vNNN.md; legacy-compatible SPEC-NN.R_review_report_vNNN.md
3. YAML Frontmatter: Required with artifact_type: SPEC-REVIEW, layer: 9
4. Score Field: tasks_ready_score_claimed / tasks_ready_score_validated
5. Parent Reference: Must link to parent SPEC document

Example Location (ALWAYS use nested folders):

# Monolithic SPEC (<20k tokens):
docs/09_SPEC/SPEC-03_f3_observability/
├── SPEC-03_f3_observability.yaml        # ← Single YAML file
├── SPEC-03.A_audit_report_v001.md       # ← Preferred combined audit report
├── SPEC-03.F_fix_report_v001.md         # ← Fix report
└── .drift_cache.json                     # ← Drift cache

# Split SPEC (≥20k tokens or multiple components):
docs/09_SPEC/SPEC-01_f1_iam/
├── SPEC-01.01_authentication.yaml       # ← Component 1
├── SPEC-01.02_authorization.yaml        # ← Component 2
├── SPEC-01.03_user_profile.yaml         # ← Component 3
├── SPEC-01.A_audit_report_v001.md       # ← Preferred combined audit report
├── SPEC-01.F_fix_report_v001.md         # ← Fix report
└── .drift_cache.json                     # ← Drift cache

Nested Folder Rule: ALL SPECs use nested folders (SPEC-NN_{slug}/) regardless of size. This keeps YAML files and companion files (review reports, fix reports, drift cache) organized together.

Audit Wrapper Compatibility: doc-spec-audit may emit preferred SPEC-NN.A_audit_report_vNNN.md; reviewer output SPEC-NN.R_review_report_vNNN.md remains valid legacy input for fixer.

Version History