From sandboxxer
Use when user wants to audit sandbox security, review firewall configuration, check for security best practices, or harden their Claude Code Docker sandbox environment
How this skill is triggered — by the user, by Claude, or both
Slash command
/sandboxxer:sandboxxer-auditThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Performs comprehensive security audits of Claude Code Docker sandbox configurations and provides recommendations for hardening based on best practices.
Performs comprehensive security audits of Claude Code Docker sandbox configurations and provides recommendations for hardening based on best practices.
Use this skill when:
Do NOT use this skill when:
/sandboxxer:quickstart)/sandboxxer:troubleshoot)/sandboxxer:linux-troubleshoot)Via slash command:
/sandboxxer:audit
Via natural language:
The skill will:
12-step security audit workflow with findings categorization and actionable recommendations.
User: "Can you audit my sandbox security?"
Assistant: "I'll perform a comprehensive security audit of your Claude Code Docker sandbox."
The skill will systematically check:
The audit provides:
Check for security issues in:
.devcontainer/devcontainer.json.devcontainer/Dockerfile.devcontainer/init-firewall.shdocker-compose.ymlCheck Firewall Mode:
grep FIREWALL_MODE .devcontainer/devcontainer.json
echo $FIREWALL_MODE
Verify Allowed Domains:
ALLOWED_DOMAINS array in init-firewall.shRecommendations:
Report Format:
Firewall Audit:
- Mode: [strict/permissive]
- Whitelisted domains: [count]
- Concerns:
⚠ domain.com - Explain why this might be problematic
✓ api.anthropic.com - Necessary for Claude Code
Check for Hardcoded Credentials:
# Search for common password patterns
grep -r "password.*=" .devcontainer/ docker-compose.yml
grep -r "API_KEY.*=" .devcontainer/
grep -r "SECRET.*=" .devcontainer/
Verify Default Passwords:
docker-compose.yml for devpassword, rootpassword, etc..env files are in .gitignoreRecommendations:
${localEnv:API_KEY}.env files (and add to .gitignore)Check Exposed Ports in docker-compose.yml:
# BAD: Unnecessary exposure
postgres:
ports:
- "5432:5432" # Only needed if accessing from host
# GOOD: Internal only
postgres:
# No ports section - only accessible from Docker network
Recommendations:
Check User Configuration in Dockerfile:
# GOOD: Non-root user
USER node
# BAD: Running as root
# USER root
Check Linux Capabilities:
"runArgs": [
"--cap-add=NET_ADMIN", // Required for firewall
"--cap-add=NET_RAW" // Required for packet filtering
]
Recommendations:
Check Volume Mounts:
"workspaceMount": "source=${localWorkspaceFolder},target=/workspace,type=bind"
Recommendations:
Check Network Configuration:
networks:
default:
name: my-project-network
Verify:
For Python projects:
# Check for known vulnerabilities
uv add safety
safety check
For Node.js projects:
npm audit
npm audit fix
Recommendations:
Check Lifecycle Commands in devcontainer.json:
initializeCommand - ⚠️ Runs on HOST before container startsonCreateCommand - Runs once when container createdupdateContentCommand - Runs on rebuildpostCreateCommand - Runs after creationpostStartCommand - Runs on every startpostAttachCommand - Runs on every attachExample Configuration:
{
"initializeCommand": ["docker", "version"],
"onCreateCommand": "npm install",
"postCreateCommand": "npm run build",
"postAttachCommand": "echo 'Container attached'"
}
Security Concerns:
initializeCommand has HOST access - audit carefullyReference: Lifecycle Scripts
Check Features in devcontainer.json:
"features": {
"ghcr.io/devcontainers/features/node:1": {
"version": "18"
},
"ghcr.io/devcontainers/features/python:1": {}
}
Security Recommendations:
ghcr.io/devcontainers/features/*:1 not :latest)Official Features Registry: Over 200 pre-built features available at Dev Container Features
Reference: Dev Container Features Specification
Check Dotfiles Configuration in devcontainer.json:
{
"dotfilesRepository": "your-github-user/dotfiles",
"dotfilesInstallCommand": "install.sh",
"dotfilesTargetPath": "~/dotfiles"
}
Security Concerns:
Reference: Personalizing with Dotfiles
Types of Environment Variables in devcontainer.json:
{
"containerEnv": {
"NODE_ENV": "development"
},
"remoteEnv": {
"DISPLAY": "${localEnv:DISPLAY}"
}
}
Environment Variable Patterns:
containerEnv - Available in containerremoteEnv - Available to VS Code processes${localEnv:VAR} - Forwarded from host.env files - Loaded from workspace rootSecurity Concerns:
containerEnv${localEnv:API_KEY} for secrets.env files should be in .gitignoreGitHub Codespaces Integration:
Reference: Variables in devcontainer.json
Provide comprehensive report:
# Security Audit Report - [Project Name]
## Summary
- Overall Risk Level: [Low/Medium/High]
- Critical Issues: [count]
- Warnings: [count]
- Recommendations: [count]
## Critical Issues ❌
1. [Issue] - [Explanation] - [Fix]
## Warnings ⚠
1. [Warning] - [Explanation] - [Recommendation]
## Good Practices ✅
1. [What's done well]
## Recommendations
1. [Improvement] - [Why] - [How]
## Security Checklist
- [ ] Firewall configured and tested
- [ ] No hardcoded credentials
- [ ] Default passwords changed/not in production
- [ ] Minimal port exposure
- [ ] Non-root user configured
- [ ] Dependencies audited
- [ ] Secrets properly managed
- [ ] Network isolation verified
- [ ] Lifecycle hooks reviewed (especially initializeCommand)
- [ ] Dev Container features from trusted sources
- [ ] Feature versions pinned
- [ ] Dotfiles repository trusted and reviewed
- [ ] Environment variables not exposing secrets
- [ ] GitHub Codespaces secrets configured properly (if using Codespaces)
From docs/features/SECURITY-MODEL.md:
What we protect against:
What we DON'T protect against:
User requests a security review of their existing DevContainer:
User: Can you audit my sandbox security?
Assistant: I'll run the security auditor to review your configuration.
/sandboxxer:audit
The skill scans configuration files, checks firewall settings, and provides a prioritized list of security recommendations.
User preparing to deploy and wants to ensure security hardening:
User: We're about to go live. Can you check if our development environment is secure?
Assistant: I'll perform a comprehensive security audit focusing on production readiness.
/sandboxxer:audit
The skill performs deep security analysis, reviews all network configurations, and provides deployment-specific recommendations.
User experiencing connectivity issues and wants to verify firewall is properly configured:
User: My API calls are being blocked. Can you check my firewall?
Assistant: I'll audit your firewall configuration.
/sandboxxer:audit
The skill reviews firewall mode, allowed domains, and provides guidance on adding necessary domains while maintaining security.
npx claudepluginhub andrewcchoi/sandbox-maxxing --plugin sandboxxerProvides Docker security guidelines covering image selection, minimal bases, build/runtime protections, network isolation, least privilege, and defense-in-depth strategies. Useful for securing containers.
Hardens Docker/container images and runtime deployments with secure base images, non-root users, CVE scanning, SBOM/signing, seccomp/AppArmor, and Kubernetes pod security controls.
Sets up and launches a Docker devcontainer running Claude Code with --dangerously-skip-permissions for autonomous sandboxed coding without prompts. Triggers on 'yolo' or 'autonomous mode'.