# Security Best Practices
This document outlines security best practices for configuring and using the Windows CLI MCP Server safely.
## Core Security Principles
### Defense in Depth
Implement multiple layers of security controls:
1. **Path Restrictions** - Limit accessible directories
2. **Command Blocking** - Block dangerous operations
3. **Input Validation** - Prevent injection attacks
4. **Process Controls** - Limit execution time and resources
5. **Audit Logging** - Monitor and review activities
### Principle of Least Privilege
- Grant only the minimum permissions necessary
- Start with restrictive settings and relax as needed
- Regularly review and tighten permissions
- Use different security levels for different environments
## Environment-Specific Configurations
### Production Environment
**Ultra-restrictive configuration for production systems:**
```json
{
"global": {
"security": {
"maxCommandLength": 200,
"commandTimeout": 5,
"enableInjectionProtection": true,
"restrictWorkingDirectory": true
},
"restrictions": {
"blockedCommands": [
// System commands
"format", "fdisk", "diskpart", "bcdedit",
"shutdown", "restart", "reboot", "halt",
// File operations
"del", "rm", "rmdir", "rd", "erase",
"move", "mv", "copy", "cp",
// Registry/System
"reg", "regedit", "regedt32",
"sc", "net", "netsh", "wmic",
// Privileges
"runas", "sudo", "su", "takeown", "icacls",
// Shells (prevent shell escape)
"cmd", "powershell", "pwsh", "bash", "sh",
// Package managers
"npm", "pip", "chocolatey", "winget",
// Compilers/Interpreters
"python", "node", "javac", "gcc", "cl"
],
"blockedArguments": [
"--exec", "-e", "/c", "-c", "-command",
"--interactive", "-i", "--login", "-l",
"--system", "-s", "--force", "-f",
"--recursive", "-r", "-rf", "--all",
"--privileged", "--admin", "--elevated"
],
"blockedOperators": [
"&", "&&", "|", "||", ";", "`", "$(",
">", ">>", "<", "<<", "^", "!", "*"
]
},
"paths": {
"allowedPaths": [
"C:\\SafeScripts",
"C:\\ApprovedData"
],
"initialDir": "C:\\SafeScripts"
}
},
"shells": {
"cmd": {
"enabled": true,
"executable": {
"command": "cmd.exe",
"args": ["/c"]
},
"overrides": {
"security": {
"maxCommandLength": 100
},
"restrictions": {
"allowedCommands": [
"echo Safe message",
"type C:\\ApprovedData\\*.txt",
"dir C:\\SafeScripts"
]
}
}
}
}
}
```
### Development Environment
**Balanced security for development work:**
```json
{
"global": {
"security": {
"maxCommandLength": 3000,
"commandTimeout": 120,
"enableInjectionProtection": true,
"restrictWorkingDirectory": true
},
"restrictions": {
"blockedCommands": [
// Critical system commands only
"format", "shutdown", "restart",
"reg", "regedit", "net user",
"takeown", "icacls"
],
"blockedArguments": [
"--system", "--privileged", "--admin"
],
"blockedOperators": [
"&", "|", ";", "`"
]
},
"paths": {
"allowedPaths": [
"C:\\Development",
"C:\\Users\\Developer\\Projects",
"C:\\temp"
],
"initialDir": "C:\\Development"
}
}
}
```
### Testing Environment
**Restrictive but functional for testing:**
```json
{
"global": {
"security": {
"maxCommandLength": 1500,
"commandTimeout": 60,
"enableInjectionProtection": true,
"restrictWorkingDirectory": true
},
"restrictions": {
"blockedCommands": [
"format", "shutdown", "restart", "reboot",
"reg", "regedit", "net", "netsh",
"takeown", "icacls", "del", "rm"
],
"blockedOperators": ["&", "|", ";", "`", "$("]
},
"paths": {
"allowedPaths": [
"C:\\TestEnvironment",
"C:\\TestData",
"C:\\TestResults"
]
}
}
}
```
## Shell-Specific Security
### PowerShell Security
PowerShell requires special attention due to its power:
```json
{
"shells": {
"powershell": {
"enabled": true,
"executable": {
"command": "powershell.exe",
"args": [
"-NoProfile", // Don't load user profile
"-NonInteractive", // Prevent interactive prompts
"-ExecutionPolicy", "Restricted", // Block script execution
"-Command"
]
},
"overrides": {
"restrictions": {
"blockedCommands": [
// PowerShell-specific dangerous commands
"Invoke-Expression", "iex", "Invoke-Command",
"New-Object", "Add-Type", "Invoke-RestMethod",
"Start-Process", "Stop-Process", "Remove-Item",
"Set-ExecutionPolicy", "Import-Module",
"Enable-PSRemoting", "Invoke-WmiMethod"
],
"blockedArguments": [
"-EncodedCommand", "-WindowStyle", "Hidden",
"-ExecutionPolicy", "Bypass", "-File",
"-ScriptBlock", "-FilterScript"
]
}
}
}
}
}
```
### CMD Security
CMD is generally safer but still needs restrictions:
```json
{
"shells": {
"cmd": {
"enabled": true,
"executable": {
"command": "cmd.exe",
"args": ["/c"]
},
"overrides": {
"restrictions": {
"blockedCommands": [
// CMD-specific dangerous commands
"del", "erase", "rd", "rmdir",
"format", "fdisk", "attrib",
"cacls", "takeown", "subst"
],
"blockedArguments": [
"/s", "/q", "/f", "/y", // Silent/force flags
"/c", "/k" // Command execution flags
]
}
}
}
}
}
```
### WSL Security
WSL requires Unix-focused security:
```json
{
"shells": {
"wsl": {
"enabled": true,
"executable": {
"command": "wsl.exe",
"args": ["-e", "bash", "-c"]
},
"wslConfig": {
"mountPoint": "/mnt/",
"inheritGlobalPaths": false // Explicit path control
},
"overrides": {
"restrictions": {
"blockedCommands": [
// Unix dangerous commands
"rm", "rmdir", "dd", "mkfs",
"sudo", "su", "passwd", "chown", "chmod",
"mount", "umount", "fdisk", "parted",
"iptables", "ufw", "systemctl", "service"
],
"blockedArguments": [
"-rf", "--recursive", "--force",
"--privileged", "--root", "--system"
]
},
"paths": {
"allowedPaths": [
"/home/user/safe",
"/tmp/workspace"
]
}
}
}
}
}
```
### Git Bash Security
Git Bash combines Windows and Unix risks:
```json
{
"shells": {
"gitbash": {
"enabled": true,
"executable": {
"command": "C:\\Program Files\\Git\\bin\\bash.exe",
"args": ["-c"]
},
"overrides": {
"restrictions": {
"blockedCommands": [
// Both Windows and Unix commands
"rm", "del", "rmdir", "rd",
"chmod", "chown", "sudo", "su"
],
"blockedOperators": [
"&", "&&", "|", "||", ";", "`", "$(",
">", ">>", "<", "<<", "^"
]
}
}
}
}
}
```
## Path Security
### Windows Path Security
```json
{
"paths": {
"allowedPaths": [
// Safe user directories
"C:\\Users\\%USERNAME%\\Documents\\Safe",
"C:\\Users\\%USERNAME%\\Desktop\\Workspace",
// Dedicated work areas
"C:\\Workspace",
"D:\\Projects",
// Temporary areas
"C:\\Temp\\MCPWork"
],
// Avoid these dangerous paths:
// "C:\\Windows", "C:\\System32", "C:\\Program Files",
// "C:\\Users\\%USERNAME%\\AppData", "C:\\"
}
}
```
### Unix Path Security (WSL)
```json
{
"paths": {
"allowedPaths": [
// Safe user areas
"/home/user/workspace",
"/home/user/documents",
// Temporary areas
"/tmp/mcpwork",
"/var/tmp/safe",
// Mounted Windows drives (controlled)
"/mnt/c/Workspace",
"/mnt/d/Projects"
]
// Avoid these dangerous paths:
// "/", "/bin", "/sbin", "/usr/bin", "/etc",
// "/var", "/boot", "/sys", "/proc"
}
}
```
## Command Security Patterns
### Allowlist Approach (Most Secure)
Instead of blocking commands, only allow specific ones:
```json
{
"restrictions": {
"allowedCommands": [
// File viewing
"type", "cat", "less", "more",
"dir", "ls", "find",
// Text processing
"findstr", "grep", "sort", "head", "tail",
// Git operations
"git status", "git log", "git diff",
// Build tools
"npm run build", "npm test",
"dotnet build", "dotnet test"
]
}
}
```
### Blocklist Approach (More Flexible)
Block dangerous patterns while allowing others:
```json
{
"restrictions": {
"blockedCommands": [
// Administrative
"^(net|sc|wmic|reg|takeown|icacls)\\s",
// File operations
"^(del|rm|format|fdisk)\\s",
// System control
"^(shutdown|restart|reboot)\\s",
// Shells (prevent escapes)
"^(cmd|powershell|bash|sh)\\s"
]
}
}
```
## Monitoring and Auditing
### Log Analysis
Regularly review command logs for:
```bash
# Suspicious patterns
grep -E "(sudo|admin|system|exec|eval)" mcp-commands.log
# Failed commands
grep "ERROR" mcp-commands.log | tail -50
# Long-running commands
grep "timeout" mcp-commands.log
# Path access attempts
grep -E "(\\\\|/)+(windows|system|etc|bin)" mcp-commands.log
```
### Automated Monitoring
Set up alerts for:
- Commands hitting blocked lists
- Path access violations
- Timeout events
- Repeated failure patterns
- Unusual command patterns
## Incident Response
### If Security Breach Suspected
1. **Immediate Actions**:
- Stop the MCP server
- Disconnect from network if needed
- Preserve logs and evidence
2. **Investigation**:
- Review command history
- Check file system changes
- Analyze network activity
- Identify scope of access
3. **Recovery**:
- Restore from clean backups
- Update configuration
- Patch vulnerabilities
- Implement additional controls
### Hardening After Incidents
```json
{
"global": {
"security": {
"maxCommandLength": 100, // Reduce from default
"commandTimeout": 3, // Very short timeout
"enableInjectionProtection": true,
"restrictWorkingDirectory": true
},
"restrictions": {
"blockedCommands": ["*"], // Block all by default
"allowedCommands": [ // Only specific allowed commands
"echo Hello",
"dir C:\\SafeOnly"
]
}
}
}
```
## Regular Security Reviews
### Monthly Reviews
- [ ] Review allowed paths for necessity
- [ ] Check blocked command lists for completeness
- [ ] Analyze command usage patterns
- [ ] Update security configurations
- [ ] Test incident response procedures
### Quarterly Reviews
- [ ] Full security audit of configuration
- [ ] Penetration testing with known attack vectors
- [ ] Review and update documentation
- [ ] Train users on security practices
- [ ] Update threat models
### Annual Reviews
- [ ] Complete security architecture review
- [ ] Benchmark against industry standards
- [ ] Review compliance requirements
- [ ] Update security policies
- [ ] Plan security improvements
## Compliance Considerations
### PCI DSS
If handling payment data:
- Use strongest restriction levels
- Log all commands with timestamps
- Restrict network access
- Regular vulnerability scans
### HIPAA
If handling health data:
- Implement access controls
- Audit all file access
- Encrypt sensitive directories
- Control data export capabilities
### SOX
If handling financial data:
- Immutable audit logs
- Separation of duties
- Change control processes
- Regular access reviews
## Common Security Mistakes
### ❌ **DON'T DO THESE:**
```json
{
// DANGEROUS: Unrestricted paths
"paths": { "allowedPaths": ["C:\\", "/"] },
// DANGEROUS: No command restrictions
"restrictions": { "blockedCommands": [] },
// DANGEROUS: Disabled security features
"security": {
"restrictWorkingDirectory": false,
"enableInjectionProtection": false
},
// DANGEROUS: Long timeouts
"security": { "commandTimeout": 3600 }
}
```
### ✅ **DO THESE INSTEAD:**
```json
{
// SAFE: Specific allowed paths
"paths": {
"allowedPaths": ["C:\\Workspace", "/home/user/safe"]
},
// SAFE: Comprehensive restrictions
"restrictions": {
"blockedCommands": [
"format", "del", "rm", "shutdown", "reg"
]
},
// SAFE: All protections enabled
"security": {
"restrictWorkingDirectory": true,
"enableInjectionProtection": true
},
// SAFE: Short timeouts
"security": { "commandTimeout": 30 }
}
```
## Security Testing
### Test Your Configuration
```bash
# Test 1: Try accessing restricted paths
validate_directories ["C:\\Windows\\System32"]
# Test 2: Try blocked commands
execute_command { "shell": "cmd", "command": "format c:" }
# Test 3: Try injection attacks
execute_command { "shell": "cmd", "command": "dir && del *.*" }
# Test 4: Try privilege escalation
execute_command { "shell": "powershell", "command": "Start-Process cmd -Verb RunAs" }
```
### Penetration Testing Scenarios
1. **Command Injection**: Try various injection patterns
2. **Path Traversal**: Attempt to access restricted directories
3. **Shell Escape**: Try to break out to other shells
4. **Resource Exhaustion**: Test timeout and length limits
5. **Configuration Bypass**: Attempt to modify settings
Remember: Security is an ongoing process, not a one-time setup. Regularly review, test, and update your security configurations.
