qa-gate（质量保证门）

/qa-gate - Quality Gate Decision

Description

Generate a quality gate decision with persistent YAML output. Runs automated checks and optional specialist reviews, then produces a gate file that can be tracked and referenced.

Key Features:

• Persistent YAML gate files for traceability
• Standardized severity scale (low, medium, high)
• Issue ID prefixes for categorization (SEC-, PERF-, TEST-, etc.)
• WAIVED status with approval tracking
• Updates story file with gate reference
• NFR validation (security, performance, reliability, accessibility)

Usage

# Quick gate for a story (tests, types, lint only)
/qa-gate 3.1.5

# Full gate with specialist reviews
/qa-gate 3.1.5 --deep

# Gate for current branch (no story reference)
/qa-gate --branch

# Gate with specific specialists
/qa-gate 3.1.5 --security --performance

# Waive known issues
/qa-gate 3.1.5 --waive --reason "Accepted for MVP" --approved-by "Product Owner"

# Quick check without persisting file
/qa-gate 3.1.5 --dry-run

Parameters

• story - Story number (e.g., 3.1.5) or omit for branch-based review
• --deep - Run all specialist reviews (security, performance, accessibility)
• --security - Run security specialist review
• --performance - Run performance specialist review
• --accessibility - Run accessibility specialist review
• --branch - Review current branch without story reference
• --waive - Mark gate as WAIVED (requires --reason and --approved-by)
• --reason - Reason for waiver
• --approved-by - Who approved the waiver
• --dry-run - Run checks but don't persist gate file

EXECUTION INSTRUCTIONS

Phase 1: Parse Arguments & Locate Files

1. Parse story number or --branch flag
2. If story provided:
   - Find story file: docs/stories/{STORY_NUM}.*.md
   - Extract story title and slug
3. Determine gate file path: docs/qa/gates/{STORY_NUM}-{slug}.yml
4. Check if gate file already exists (for history tracking)

Phase 2: Run Required Checks

Always run these checks first:

# Run in parallel for speed
pnpm test --filter='...[origin/main]'
pnpm check-types --filter='...[origin/main]'
pnpm lint --filter='...[origin/main]'

Collect results:

checks:
  tests: { status: PASS|FAIL, details: "X tests passed" }
  types: { status: PASS|FAIL, details: "No type errors" }
  lint: { status: PASS|FAIL, details: "No lint errors" }

Phase 3: Specialist Reviews (if --deep or specific flags)

Use haiku model for fast, focused reviews:

# Security Review (if --deep or --security)
Task(
  subagent_type: "general-purpose",
  model: "haiku",
  description: "Security review",
  prompt: "You are a security specialist. Review the code changes.

           Check for:
           - Authentication/authorization issues
           - Injection vulnerabilities (SQL, XSS, command)
           - Sensitive data exposure
           - OWASP Top 10 issues
           - Hardcoded secrets or credentials

           For each finding, provide:
           - id: SEC-{NNN}
           - severity: low|medium|high
           - finding: Brief description
           - suggested_action: How to fix
           - file: File path
           - line: Line number (if applicable)

           Return findings as YAML array."
)

# Performance Review (if --deep or --performance)
Task(
  subagent_type: "general-purpose",
  model: "haiku",
  description: "Performance review",
  prompt: "You are a performance specialist. Review the code changes.

           Check for:
           - N+1 query patterns
           - Missing database indexes
           - Unnecessary re-renders in React
           - Large bundle imports
           - Missing memoization
           - Inefficient algorithms

           For each finding, provide:
           - id: PERF-{NNN}
           - severity: low|medium|high
           - finding: Brief description
           - suggested_action: How to fix
           - file: File path

           Return findings as YAML array."
)

# Accessibility Review (if --deep or --accessibility)
Task(
  subagent_type: "general-purpose",
  model: "haiku",
  description: "Accessibility review",
  prompt: "You are an accessibility specialist. Review the code changes.

           Check for:
           - WCAG 2.1 AA compliance
           - Keyboard navigation support
           - Screen reader compatibility
           - Missing ARIA labels/roles
           - Color contrast issues
           - Focus management

           For each finding, provide:
           - id: A11Y-{NNN}
           - severity: low|medium|high
           - finding: Brief description with WCAG criterion
           - suggested_action: How to fix
           - file: File path

           Return findings as YAML array."
)

Phase 4: Determine Gate Decision

Decision logic:

IF --waive flag provided:
  gate = WAIVED
ELSE IF any check FAILED OR any high severity issue:
  gate = FAIL
ELSE IF any medium severity issue:
  gate = CONCERNS
ELSE:
  gate = PASS

Status reason (1-2 sentences):

• PASS: "All checks passed with no significant issues."
• CONCERNS: "Non-blocking issues found that should be addressed."
• FAIL: "{reason for failure - e.g., 'Tests failing' or 'High severity security issue'}"
• WAIVED: "{user-provided reason}"

Phase 5: Generate Gate File

Write YAML to docs/qa/gates/{STORY_NUM}-{slug}.yml:

schema: 1
story: "{STORY_NUM}"
story_title: "{STORY_TITLE}"
gate: PASS|CONCERNS|FAIL|WAIVED
status_reason: "{1-2 sentence explanation}"
reviewer: "Claude Code"
updated: "{ISO-8601 timestamp}"

# Waiver (only active if WAIVED)
waiver:
  active: false  # or true if WAIVED
  reason: ""     # populated if WAIVED
  approved_by: "" # populated if WAIVED

# All issues found
top_issues: []
# Example:
#   - id: "SEC-001"
#     severity: high
#     finding: "No rate limiting on login endpoint"
#     suggested_action: "Add rate limiting middleware"
#     file: "src/api/auth/login.ts"

# NFR validation summary
nfr_validation:
  security: { status: PASS|CONCERNS|FAIL|SKIPPED, issue_count: 0 }
  performance: { status: PASS|CONCERNS|FAIL|SKIPPED, issue_count: 0 }
  accessibility: { status: PASS|CONCERNS|FAIL|SKIPPED, issue_count: 0 }
  tests: { status: PASS|FAIL, details: "" }
  types: { status: PASS|FAIL, details: "" }
  lint: { status: PASS|FAIL, details: "" }

# Risk summary
risk_summary:
  totals: { high: 0, medium: 0, low: 0 }
  recommendations:
    must_fix: []   # high severity items
    should_fix: [] # medium severity items

Phase 6: Update Story File (if story provided)

Append to story's QA Results section:

## QA Results

### Gate Status

Gate: {STATUS} → docs/qa/gates/{STORY_NUM}-{slug}.yml
Updated: {ISO-8601 timestamp}
Reviewer: Claude Code

{If issues found:}
Top Issues:
- [{ID}] {severity}: {finding}

Phase 7: Report Summary

═══════════════════════════════════════════════════════
  Quality Gate: {STORY_NUM} - {STORY_TITLE}
═══════════════════════════════════════════════════════

Gate: {PASS|CONCERNS|FAIL|WAIVED}
Reason: {status_reason}

Checks:
  Tests:    {PASS|FAIL}
  Types:    {PASS|FAIL}
  Lint:     {PASS|FAIL}

{If specialist reviews run:}
NFR Validation:
  Security:      {STATUS} ({N} issues)
  Performance:   {STATUS} ({N} issues)
  Accessibility: {STATUS} ({N} issues)

{If issues found:}
Top Issues ({N} total):
  [{ID}] {severity}: {finding}
  ...

Gate File: docs/qa/gates/{STORY_NUM}-{slug}.yml

{If FAIL:}
Recommendation: Address high-severity issues before proceeding.

{If CONCERNS:}
Recommendation: Review issues and proceed with awareness.

{If WAIVED:}
Waiver: {reason}
Approved By: {approved_by}
═══════════════════════════════════════════════════════

Issue ID Prefixes

Severity Scale

Fixed values - no variations:

Gate Statuses

Sub-Agent Architecture

Main Orchestrator (/qa-gate)
    │
    ├─▶ Required Checks (inline)
    │   ├── pnpm test
    │   ├── pnpm check-types
    │   └── pnpm lint
    │
    └─▶ Specialist Reviews (parallel, haiku)
        ├── Security specialist
        ├── Performance specialist
        └── Accessibility specialist

Integration with /implement

The /implement skill calls /qa-gate for its QA phase:

/implement 3.1.5
    │
    ├─▶ ... implementation phases ...
    │
    └─▶ /qa-gate 3.1.5 --deep
            └── Produces gate file and updates story

Examples

Quick Gate (tests only)

/qa-gate 3.1.5
# Runs: tests, types, lint
# Output: docs/qa/gates/3.1.5-my-story.yml

Deep Gate (all specialists)

/qa-gate 3.1.5 --deep
# Runs: tests, types, lint + security, performance, accessibility
# Output: docs/qa/gates/3.1.5-my-story.yml

Waive Known Issues

/qa-gate 3.1.5 --waive --reason "MVP release, will fix in v2" --approved-by "Tech Lead"
# Sets gate to WAIVED with approval tracking

Branch Review (no story)

/qa-gate --branch --deep
# Reviews current branch without story reference
# Output: docs/qa/gates/branch-{branch-name}.yml