Skip to content
Back to skills

Code Review Gate

ASecurity

Human code review gate for critical components. Activates automatically for security-sensitive paths, high-complexity code, or large changes. Activation trigger: [ACTIVATE:CODE_REVIEW_GATE_V1]

  • 666 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added February 8, 2026
code-qualitygobashsqltestingcode-reviewgitapisecuritydocumentation

Works with

  • api

Security analysis

A100/100

Scanned February 12, 2026

npx -y skills add majiayu000/claude-skill-registry --skill code-review-gate --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Code Review Gate?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Code Review Gate
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/majiayu000-code-review-gate/badge)](https://www.skillsdirectory.com/skills/majiayu000-code-review-gate)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
activation_code: CODE_REVIEW_GATE_V1
phase: 8
prerequisites:
  - Phase 7 implementation complete
  - Tests passing
  - Coverage thresholds met
outputs:
  - Code review report
  - .signals/code-review-complete.json
description: |
  Human code review gate for critical components. Activates automatically
  for security-sensitive paths, high-complexity code, or large changes.

  Activation trigger: [ACTIVATE:CODE_REVIEW_GATE_V1]
---

# Code Review Gate Skill

## Purpose

This skill provides a **human code review gate** between implementation (Phase 7) and integration testing (Phase 9). It ensures that code quality, security, and maintainability are validated by human eyes before proceeding.

## When This Gate Triggers

### Automatic Triggers (Cannot Skip)

1. **Security-Sensitive Paths**
   - `src/auth/` - Authentication logic
   - `src/security/` - Security utilities
   - `src/crypto/` - Cryptographic operations
   - `api/` - API endpoints
   - `services/` - External service integrations

2. **High Complexity Code**
   - Any function with cyclomatic complexity > 15
   - Files with > 500 lines
   - Functions with > 50 lines

3. **Large Changes**
   - > 10 files modified in a phase
   - > 500 lines added/modified

### Manual Triggers

- User says "review code" or "code review"
- User says "check my implementation"
- Activated via `/code-review` command

## Review Checklist

### Security Review

- [ ] No hardcoded secrets (API keys, passwords, tokens)
- [ ] Input validation on all external inputs
- [ ] Output encoding to prevent XSS
- [ ] SQL queries use parameterized statements
- [ ] Authentication checks on protected routes
- [ ] Authorization checks for resource access
- [ ] Sensitive data not logged
- [ ] HTTPS enforced for external calls

### Code Quality Review

- [ ] Functions have single responsibility
- [ ] No code duplication (DRY)
- [ ] Meaningful variable/function names
- [ ] Complex logic has comments explaining WHY
- [ ] Error handling is appropriate
- [ ] No silent failures (catch without action)
- [ ] Resource cleanup (files, connections, etc.)

### Architecture Review

- [ ] Follows layer boundaries (L0-L5)
- [ ] Dependencies flow downward only
- [ ] Interface contracts respected
- [ ] No circular dependencies
- [ ] Separation of concerns maintained

### Test Coverage Review

- [ ] Critical paths have 100% coverage
- [ ] Edge cases are tested
- [ ] Error paths are tested
- [ ] Assertions are meaningful (not just execution)
- [ ] No mocks in production code

### Documentation Review

- [ ] Public functions have docstrings/JSDoc
- [ ] Complex algorithms are explained
- [ ] API changes documented
- [ ] README updated if needed

## Review Process

### Step 1: Gather Changes

```bash
# Show all files changed in this phase
git diff --stat HEAD~10..HEAD

# Show detailed diff
git diff HEAD~10..HEAD

# List files by complexity (if radon available)
radon cc -s src/
```

### Step 2: Analyze Risk

The skill analyzes changes and categorizes them:

| Category | Action | Blocking |
|----------|--------|----------|
| Security-critical | Must review all changes | YES |
| High-complexity | Must review flagged functions | YES |
| API changes | Must review contracts | YES |
| Internal changes | Spot-check recommended | NO |

### Step 3: Present to Human

```
═══════════════════════════════════════════════════════════════
  CODE REVIEW REQUIRED
═══════════════════════════════════════════════════════════════

Trigger: Security-sensitive files modified

Files for Review:
  1. src/auth/login.py (52 lines added)
     - New authentication logic
     - Complexity: 8/10
     - Security: HIGH PRIORITY

  2. src/auth/tokens.py (34 lines added)
     - JWT token handling
     - Complexity: 5/10
     - Security: HIGH PRIORITY

  3. api/users.py (78 lines added)
     - User CRUD operations
     - Complexity: 6/10
     - Security: MEDIUM PRIORITY

Review Checklist:
  [ ] Security review completed
  [ ] Code quality acceptable
  [ ] Architecture boundaries respected
  [ ] Test coverage verified

Actions:
  - "approve" - Proceed to Phase 6
  - "reject [reason]" - Return to implementation
  - "comment [file:line] [note]" - Add review comment
  - "show [file]" - Display file contents
  - "diff [file]" - Show file diff
═══════════════════════════════════════════════════════════════
```

### Step 4: Human Decision

**Approve:**
```
approve

✅ CODE REVIEW APPROVED
Proceeding to Phase 9 (Integration Testing)
[SIGNAL:CODE_REVIEW_COMPLETE]
[ACTIVATE:INTEGRATION_VALIDATOR_V1]
```

**Reject:**
```
reject Security: Missing input validation on user_id parameter in api/users.py:45

❌ CODE REVIEW REJECTED
Reason: Security: Missing input validation on user_id parameter in api/users.py:45

Returning to Phase 7 for fixes.
Please address the review feedback and re-submit.
```

**Add Comment:**
```
comment src/auth/login.py:23 Consider using constant-time comparison for password check

📝 Comment added to review log
Continue with: approve | reject | more comments
```

## Output Files

### Review Report

`.claude/reports/code-review-report.json`:
```json
{
  "phase": 5.5,
  "timestamp": "2025-12-19T12:00:00Z",
  "status": "approved",
  "reviewer": "human",
  "trigger": "security_sensitive_paths",
  "files_reviewed": 3,
  "security_items": 2,
  "complexity_items": 1,
  "comments": [
    {
      "file": "src/auth/login.py",
      "line": 23,
      "comment": "Consider using constant-time comparison for password check"
    }
  ],
  "checklist": {
    "security": true,
    "code_quality": true,
    "architecture": true,
    "test_coverage": true,
    "documentation": true
  }
}
```

### Completion Signal

`.claude/.signals/code-review-complete.json`:
```json
{
  "phase": 8,
  "status": "approved",
  "timestamp": "2025-12-19T12:00:00Z",
  "next_phase": 9,
  "trigger_next": true
}
```

## Configuration

In `config/quality-rules.json`:

```json
{
  "code_review_gate": {
    "enabled": true,
    "trigger_on": {
      "security_sensitive_paths": [
        "src/auth/",
        "src/security/",
        "api/"
      ],
      "high_complexity_threshold": 15,
      "file_changes_threshold": 10
    }
  }
}
```

To disable for a project:
```json
{
  "code_review_gate": {
    "enabled": false
  }
}
```

## Bypass (Emergency Only)

For emergency situations only, the gate can be bypassed:

```
bypass-review --reason "Production hotfix for CVE-2025-XXXXX" --ticket JIRA-1234
```

Bypass requirements:
- Must provide reason
- Must provide ticket/tracking number
- Logged to audit trail
- Triggers post-deployment review

## Integration with Pipeline

```
Phase 7: TDD Implementation
    ↓
Phase 8: Code Review Gate (NEW)
    ├─ Automatic for security/complexity triggers
    ├─ Human reviews code
    └─ Approve/Reject decision
    ↓
Phase 9: Integration Testing
```

## See Also

- `hooks/code-quality-validator.sh` - Automated quality checks
- `config/quality-rules.json` - Quality thresholds
- `skills/prd-audit/SKILL.md` - PRD quality audit

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.