AIES
PortalAbout
Docs/03 skill engineering/skills/security skill

Security Review Skill

Skill ID: security Version: 1.0 Install: .claude/skills/security.skill.md

Trigger Conditions

Activate this skill when:

  • Security review is explicitly requested
  • A PR touches authentication, authorization, or data access
  • New API endpoints are being added
  • External data is being processed
  • Cryptography, tokens, or secrets are involved
  • User data (PII, financial, health) is being handled

Purpose

Perform a comprehensive security review against OWASP Top 10 and AI-specific attack vectors. Produce a prioritized finding report with actionable remediation.

Prerequisites

  • Access to the code under review
  • Understanding of the application's trust boundaries
  • Knowledge of data sensitivity levels (public, internal, confidential, restricted)

Procedure

Step 1: Map the Attack Surface

Identify:

  • All entry points (API endpoints, webhooks, file uploads, CLI inputs)
  • All data stores (databases, caches, file systems, external APIs)
  • All trust boundaries (public internet, internal network, third-party integrations)
  • All privilege levels (anonymous, authenticated, admin, system)

Step 2: OWASP Top 10 Audit

A01: Broken Access Control

  • Authorization checked on every protected route/resource
  • Horizontal privilege escalation tested (can user A access user B's data?)
  • Vertical privilege escalation tested (can regular user access admin functions?)
  • IDOR (Insecure Direct Object Reference) — are database IDs exposed in URLs?

A02: Cryptographic Failures

  • Passwords hashed with bcrypt/Argon2 (NOT MD5, SHA1, or SHA256 alone)
  • Sensitive data encrypted at rest and in transit
  • TLS enforced on all endpoints (no HTTP fallback)
  • No hardcoded secrets in source code
  • Secrets loaded from environment variables / secret manager

A03: Injection

  • SQL injection: All queries use parameterized statements or ORM
  • NoSQL injection: Input validated before MongoDB/Redis operations
  • Command injection: Shell commands avoid user input; use subprocess arrays
  • XSS: All user content sanitized/escaped before rendering
  • Path traversal: File paths validated against allowlist

A04: Insecure Design

  • Threat model documented for sensitive features
  • Rate limiting on authentication endpoints
  • Account lockout after failed attempts
  • Sensitive operations require re-authentication

A05: Security Misconfiguration

  • Default credentials changed
  • Error messages do not expose stack traces or internal structure
  • Debug modes disabled in production
  • CORS properly configured (not * in production)
  • Security headers present (CSP, HSTS, X-Frame-Options)

A06: Vulnerable Components

  • No known CVEs in direct dependencies (run npm audit, pip-audit, etc.)
  • Dependencies up to date or with documented exceptions

A07: Authentication Failures

  • Session tokens cryptographically random (128+ bits)
  • Session invalidated on logout
  • Password reset tokens time-limited and single-use
  • MFA available for sensitive operations

A08: Software Integrity Failures

  • Dependency integrity verified (lock files committed)
  • No untrusted code executed at runtime

A09: Security Logging

  • Authentication events logged (success and failure)
  • Authorization failures logged
  • Sensitive data access logged
  • Logs do not contain sensitive data (passwords, tokens, PII)

A10: SSRF

  • Outbound HTTP requests validate destination against allowlist
  • Internal network addresses blocked from user-controlled URLs
  • Redirects followed with validation

Step 3: AI-Specific Security (If Applicable)

  • Prompt injection: User input separated from system instructions via XML tags
  • Tool execution: Principle of least privilege — tools cannot exceed user's permissions
  • Context isolation: Multi-tenant systems filter context by tenant at database level
  • Output validation: Model output validated before acting on it
  • Audit trail: All AI actions logged with user, model, tokens, action taken

Step 4: Prioritize Findings

Assign severity using CVSS-inspired framework:

  • Critical: Remote code execution, authentication bypass, mass data exposure
  • High: Privilege escalation, SQL injection, SSRF
  • Medium: XSS, CSRF, information disclosure
  • Low: Missing security headers, verbose errors, minor information leaks
  • Informational: Defense-in-depth improvements, hardening recommendations

Output Format

## Security Review Report

**Scope:** [Files/systems reviewed]
**Date:** [YYYY-MM-DD]
**Reviewer:** Security Review Skill v1.0

### Attack Surface Map
[Brief description of entry points, data stores, trust boundaries]

### Findings

#### CRITICAL
| ID | Finding | Location | Remediation |
|----|---------|----------|-------------|
| S-001 | [Finding] | [File:Line] | [Fix] |

#### HIGH
[Same table format]

#### MEDIUM
[Same table format]

#### LOW / INFORMATIONAL
[Same table format]

### Not Reviewed
[Items explicitly out of scope and why]

### Recommended Next Steps
1. [Immediate action — within 24 hours]
2. [Short-term — within 1 sprint]
3. [Medium-term — within 1 quarter]

Verification

Security review succeeded when:

  • All OWASP Top 10 categories addressed
  • All Critical and High findings have remediation steps
  • Report shared with team before merge
  • At least one finding has a test case to prevent regression

Do Not Use This Skill When

  • Reviewing infrastructure or network security (use a dedicated network security tool)
  • Conducting a full penetration test (this skill is a code review, not a pentest)
  • The code is clearly not security-sensitive (pure UI components with no data access)
Version: AIES v1.0.0✏️ Edit this page on GitHub
Next ChapterStart here →