Claude Code Plugins

Community-maintained marketplace

Feedback

This skill should be used when the user says \"security audit\", \"check for vulnerabilities\", \"security review\", \"harden project\", \"dependency audit\", \"credential scan\", \"check for secrets\", \"scan for secrets\", \"OWASP review\", \"security checklist\", \"audit dependencies\", \"find vulnerabilities\", or wants to review their project for security issues, exposed credentials, or vulnerable dependencies.

Install Skill

Shared

Installs to .agents/skills, used by Codex, Amp, Warp, Cursor, OpenCode, and more.

CodexAmp
Warp
CursorOpenCode
Cline
Gemini CLI
GitHub Copilot
Personal

Available across projects.

$npx skills-installer add @neuromechanist/research-skills/security-audit --client shared
Project

Writes to .agents/skills.

$npx skills-installer add @neuromechanist/research-skills/security-audit -p --client shared
Note: Review the skill instructions before using it.

SKILL.md

name security-audit
description This skill should be used when the user says "security audit", "check for vulnerabilities", "security review", "harden project", "dependency audit", "credential scan", "check for secrets", "scan for secrets", "OWASP review", "security checklist", "audit dependencies", "find vulnerabilities", or wants to review their project for security issues, exposed credentials, or vulnerable dependencies.
version 0.1.0

Security Audit

Systematic security review of a project covering dependency vulnerabilities, credential exposure, common code vulnerabilities, and configuration hardening.

When to Use

  • Before a release or deployment
  • After adding new dependencies
  • When onboarding to a new codebase
  • Periodic security reviews
  • After receiving a vulnerability report

Audit Checklist

1. Credential and Secret Scanning

Check for exposed secrets in the codebase:

# Check for common secret patterns in tracked files
git grep -n -i -E '(api_key|apikey|secret|password|token|credential|private_key)\s*[:=]' -- ':!*.md' ':!*.lock'

# Check for .env files tracked in git
git ls-files | grep -i '\.env'

# Check .gitignore covers sensitive files
for f in .env .env.local credentials.json secrets.yaml; do
  git check-ignore "$f" 2>/dev/null || echo "WARNING: $f not in .gitignore"
done

Files that must never be committed:

  • .env, .env.* (environment variables)
  • credentials.json, service-account.json (cloud credentials)
  • *.pem, *.key (private keys)
  • *.p12, *.pfx (certificates with private keys)

2. Dependency Vulnerability Scan

Python:

uv run pip-audit

JavaScript/TypeScript:

bun pm audit
# or check with npm for broader database
npm audit --omit=dev

Go:

govulncheck ./...

Review results against the severity rubric (shared with the dependency-auditor agent):

  • Critical: exploitable now or actively exploited (leaked live credential, known-exploited CVE in a reachable path). Must fix before release; blocks release-prep.
  • High: known vulnerability with a plausible attack path. Fix before the next release.
  • Medium: vulnerability with mitigating factors, or a major-version lag on a security-relevant package. Fix within the sprint.
  • Low: outdated without known vulnerabilities; license concerns to review. Track in backlog.

3. Code Vulnerability Patterns

Scan for common vulnerability patterns:

SQL Injection:

# Look for string interpolation in SQL
grep -rn 'f".*SELECT\|f".*INSERT\|f".*UPDATE\|f".*DELETE' --include='*.py'
grep -rn "format.*SELECT\|format.*INSERT" --include='*.py'

Command Injection:

# Look for shell=True or unsanitized subprocess calls
grep -rn 'shell=True\|os\.system\|subprocess\.call.*shell' --include='*.py'
grep -rn 'exec(\|eval(' --include='*.py' --include='*.js' --include='*.ts'

XSS (Cross-Site Scripting):

# Look for dangerouslySetInnerHTML or unescaped output
grep -rn 'dangerouslySetInnerHTML\|innerHTML\s*=' --include='*.tsx' --include='*.jsx' --include='*.ts' --include='*.js'

Path Traversal:

grep -rn 'open(.*\+\|os\.path\.join.*input\|req\.\(params\|query\|body\)' --include='*.py' --include='*.js'

4. Authentication and Authorization

Review:

  • Authentication endpoints use constant-time comparison for secrets
  • Session tokens have appropriate expiration
  • CORS configuration is restrictive (not * in production)
  • Rate limiting on auth endpoints
  • Password hashing uses bcrypt/argon2 (not MD5/SHA1)

5. Configuration Hardening

  • Debug mode disabled in production configs
  • HTTPS enforced (HSTS headers)
  • Security headers present (CSP, X-Frame-Options, X-Content-Type-Options)
  • Error messages do not leak stack traces in production
  • Logging does not include sensitive data (passwords, tokens)

6. Docker Security (if applicable)

  • Non-root user in Dockerfile
  • Base image is pinned to specific version (not latest)
  • No secrets in Docker build args or layers
  • Health checks configured
  • Read-only filesystem where possible

7. CI/CD Security

  • Secrets stored in GitHub Secrets (not in workflow files)
  • Third-party actions pinned to SHA (not tags)
  • Minimal permissions in workflow permissions: block
  • No pull_request_target with actions/checkout of PR head

Output Format

Present findings as a prioritized list:

## Security Audit Results

### Critical (must fix)
1. [CRED] API key found in src/config.py:42 - move to environment variable
2. [DEP] lodash 4.17.20 has prototype pollution (CVE-2021-23337)

### High (fix before release)
3. [CODE] SQL injection risk in src/db.py:88 - use parameterized queries

### Medium (fix within sprint)
4. [CONFIG] CORS allows * origin in production config

### Low (backlog)
5. [STYLE] Error responses include stack traces in non-debug mode

### Passed
- [x] No .env files in git
- [x] Docker runs as non-root
- [x] Dependencies up to date

Additional Resources