glincker/claude-code-marketplace/security-scanner

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Comprehensive security scanner for vulnerabilities, hardcoded secrets, and OWASP Top 10 issues

Install

npx vskill@latest install glincker/claude-code-marketplace/security-scanner

SKILL.md

Security Scanner

Comprehensive security analysis detecting hardcoded secrets, SQL injection, XSS, insecure dependencies, and OWASP Top 10 vulnerabilities.

What This Skill Does

  • Scans for hardcoded secrets (API keys, passwords, tokens)
  • Detects SQL injection vulnerabilities
  • Finds XSS attack vectors
  • Checks insecure dependencies
  • Identifies authentication/authorization flaws
  • Reviews CORS and CSP configurations
  • Checks for insecure cryptography

Instructions

Phase 1: Secret Detection

Search for common secret patterns:

bash
# API Keys
Grep for patterns:
- AWS: AKIA[0-9A-Z]{16}
- GitHub: ghp_[0-9a-zA-Z]{36}
- Slack: xox[baprs]-[0-9]{10,13}-[0-9a-zA-Z]{24,}
- Stripe: sk_live_[0-9a-zA-Z]{24}
- Private keys: -----BEGIN.*PRIVATE KEY-----
# Passwords
- password\s*=\s*["'][^"']+["']
- DB connection strings
- Hardcoded credentials

Example findings:

HIGH: Hardcoded AWS key found
File: src/config.js:12
Pattern: const AWS_KEY = "AKIAIOSFODNN7EXAMPLE"
Risk: Exposed AWS credentials
Fix: Use environment variables or secrets manager
Phase 2: Injection Vulnerabilities

SQL Injection:

python
# BAD: String concatenation
query = "SELECT * FROM users WHERE id = " + user_id
# GOOD: Parameterized query
query = "SELECT * FROM users WHERE id = ?"

Command Injection:

javascript
// BAD: Direct execution
exec(`ls ${userInput}`)
// GOOD: Sanitized input
execFile('ls', [sanitize(userInput)])

XSS Detection:

javascript
// BAD: innerHTML with user input
element.innerHTML = userInput
// GOOD: textContent or sanitize
element.textContent = userInput
// Or use DOMPurify
element.innerHTML = DOMPurify.sanitize(userInput)
Phase 3: Dependency Vulnerabilities
bash
# Check for vulnerable dependencies
npm audit
pip-audit
cargo audit
# Generate report
npm audit --json > audit-report.json
Phase 4: OWASP Top 10 Checks
  1. Broken Access Control
  • Missing authorization checks
  • Insecure direct object references
  • Exposed admin endpoints
  1. Cryptographic Failures
  • Hardcoded encryption keys
  • Weak algorithms (MD5, SHA1)
  • Unencrypted sensitive data
  1. Injection
  • SQL, NoSQL, OS command, LDAP injection
  • Unsanitized user input
  1. Insecure Design
  • Missing rate limiting
  • No CSRF protection
  • Weak session management
  1. Security Misconfiguration
  • Debug mode in production
  • Default credentials
  • Unnecessary services enabled

Scan Report Format

markdown
# Security Scan Report
**Scan Date**: 2025-01-13
**Project**: myapp
**Total Issues**: 12
## Critical (2)
1.Hardcoded AWS credentials
-File: config/aws.js:5
-Risk: Full AWS account compromise
-Fix: Use AWS IAM roles or environment variables
2.SQL Injection vulnerability
-File: api/users.js:45
-Code: `SELECT * FROM users WHERE id = ${userId}`
-Fix: Use parameterized queries
## High (5)
3.XSS vulnerability in search
-File: components/Search.jsx:23
-Risk: Arbitrary JavaScript execution
-Fix: Sanitize user input with DOMPurify
4.Missing authentication on /admin endpoint
-File: routes/admin.js
-Risk: Unauthorized access to admin functions
-Fix: Add authentication middleware
## Medium (3)
5.Weak CORS configuration
-File: server.js:10
-Issue: CORS set to *
-Fix: Whitelist specific domains
## Low (2)
6.Console.log in production
-Multiple files
-Risk: Information disclosure
-Fix: Remove or use proper logging
## Recommendations
1.Implement secrets management (AWS Secrets Manager, Vault)
2.Add input validation library (joi, express-validator)
3.Enable Content Security Policy
4.Implement rate limiting
5.Add security headers (helmet.js)

Auto-Fix Suggestions

javascript
// Before (Vulnerable)
const query = `SELECT * FROM users WHERE email = '${email}'`;
// After (Fixed)
const query = 'SELECT * FROM users WHERE email = ?';
db.execute(query, [email]);

Best Practices

  1. Never commit secrets: Use .env files (in .gitignore)
  2. Sanitize all input: Treat all user input as untrusted
  3. Use prepared statements: Prevent SQL injection
  4. Implement CSP: Mitigate XSS attacks
  5. Regular dependency updates: Fix known vulnerabilities
  6. Principle of least privilege: Minimal permissions

Tool Requirements

  • Read: Examine source code
  • Grep: Pattern matching for vulnerabilities
  • Glob: Find all relevant files
  • Bash: Run security tools (npm audit, etc.)

Examples

Example 1: Quick Scan

User: "Scan my code for security issues"

Output:

  • 3 Critical: Hardcoded secrets
  • 5 High: SQL injection, XSS
  • 2 Medium: CORS misconfiguration
  • Auto-fix suggestions provided
Example 2: Dependency Audit

User: "Check for vulnerable dependencies"

Output:

  • 15 vulnerabilities found
  • 10 fixable with npm audit fix
  • 5 require manual updates

Changelog

Version 1.0.0
  • Secret detection
  • Injection vulnerability scanning
  • OWASP Top 10 checks
  • Dependency auditing
  • Auto-fix suggestions

Author

GLINCKER Team

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/glincker/claude-code-marketplace/security-scanner/badge)](https://verified-skill.com/skills/glincker/claude-code-marketplace/security-scanner)

Something wrong with this skill or its page? Tell us.