cyhber-deploy

---

✨ What's New in v2.0

Complete rebuild with TDD methodology:

• ✅ Systematic 5-layer review enforced (no ad-hoc analysis)
• ✅ Standardized severity-tagged alerts (CRITICO/ALTO/MEDIO/BAJO)
• ✅ Proactive scope expansion beyond user request
• ✅ Token-efficient (~450 words, 60% reduction)
• ✅ Pressure-tested with time/authority/exhaustion scenarios
• ✅ Full test documentation included

Observed in manual evaluation: 6-8 categorized alerts vs 1-2 ad-hoc findings on the same test scenarios. These are qualitative comparisons, not automated benchmarks.

See FINAL-REPORT.md for the full evaluation notes.

---

🚀 Features

Systematic 5-Layer Analysis

Every security review follows this structured methodology:

1. Code Validation — Input validation, injection (SQL/XSS/Command), auth/authz, error handling
2. Dependencies — CVEs, outdated packages, unmaintained libraries, transitive dependencies
3. Secrets & PII — Hardcoded API keys, tokens, certificates, exposed personal data
4. CI/CD Pipeline — Branch restrictions, secret exposure, security gates, rollback plans
5. Infrastructure — Network exposure, IAM/RBAC permissions, encryption, security groups

No layer skipped — even "frontend-only" code triggers full review.

Standardized Alert Format

Every finding includes:

Campo	Valor
Severidad	🔴 CRITICO | 🟠 ALTO | 🟡 MEDIO | 🟢 BAJO
ID	CD-SEC-XXX (sequential)
Componente	file.js:line or resource name
Descripción	What + why it's a risk
Evidencia	Code snippet or config excerpt
Remediación	Specific fix steps

Risk Assessment Output

┌─────────────────────────────────────────────┐
│ 🔒 ESTADO DE SEGURIDAD                      │
├─────────────────────────────────────────────┤
│ Nivel de riesgo:  🔴 CRITICO                │
│ Alertas totales:  8                         │
│   • Críticas:     2                         │
│   • Altas:        3                         │
│   • Medias:       2                         │
│   • Bajas:        1                         │
├─────────────────────────────────────────────┤
│ ⚠️  RECOMENDACIÓN:                          │
│ BLOQUEAR despliegue hasta resolver          │
│ hallazgos críticos y altos.                 │
└─────────────────────────────────────────────┘

---

📦 Installation

Claude Code CLI

# Install from local directory
git clone https://github.com/DevCop95/cyhber-deploy.git
cd cyhber-deploy
cp -r skills/cyhber-deploy ~/.claude/skills/

Verify Installation

# Check skill is loaded
claude # Start Claude Code session
# Skill auto-triggers on: deploy, CI/CD, Terraform, auth, secrets, injection

---

🎯 Usage

Auto-Trigger Keywords

Skill activates when you mention:

• Deploy: staging, production, release
• CI/CD: GitHub Actions, GitLab CI, Jenkins, CircleCI
• IaC: Terraform, Kubernetes, Docker, Helm
• Cloud: AWS, GCP, Azure
• Security: auth, secrets, injection, SQL, XSS
• Review: "security review", "check this code"

Manual Invocation

@cyhber-deploy review this for production

---

📸 Examples

Example 1: SQL Injection Detection

Input:

app.post('/login', (req, res) => {
  db.query(`SELECT * FROM users WHERE email='${req.body.email}'`);
});

Output: