# Security Policy
## Supported Versions
| Version | Supported |
| ------- | ------------------ |
| 0.3.x | :white_check_mark: |
| 0.2.x | :white_check_mark: |
| 0.1.x | :white_check_mark: |
## Reporting a Vulnerability
We take security vulnerabilities seriously. If you discover a security issue, please report it responsibly.
### How to Report
**For security vulnerabilities, please use GitHub's private vulnerability reporting:**
1. Go to the [Security tab](https://github.com/palhamel/46elks-mcp-server/security) of this repository
2. Click "Report a vulnerability"
3. Provide details about the vulnerability
Alternatively, you can email security concerns to the repository maintainer (check the commit history for contact information).
### What to Include
When reporting a vulnerability, please include:
- Description of the vulnerability
- Steps to reproduce the issue
- Potential impact
- Any suggested fixes (optional)
### What to Expect
- **Acknowledgment**: We will acknowledge receipt within 48 hours
- **Updates**: We will provide updates on the status of your report
- **Resolution**: We aim to resolve critical vulnerabilities within 7 days
- **Credit**: We will credit reporters in the release notes (unless you prefer to remain anonymous)
## OWASP MCP Top 10 Compliance
This MCP server has been hardened following the [OWASP MCP Top 10](https://owasp.org/www-project-mcp-top-10/) guidelines:
| OWASP ID | Risk | Mitigation Status |
|----------|------|-------------------|
| MCP01 | Tool Poisoning | Mitigated - Credential validation on startup |
| MCP02 | Excessive Agency | Mitigated - Limited to SMS operations only |
| MCP03 | Insecure Output Handling | Mitigated - Input validation prevents injection |
| MCP04 | Tool Permission Boundaries | Mitigated - Local-only operation, no network exposure |
| MCP05 | Prompt Injection via Tools | Mitigated - Input sanitization, URL detection, rate limiting |
| MCP06 | Insecure Tool Input | Mitigated - Comprehensive validation (phone, message, sender ID, message ID) |
| MCP07 | Third-Party Tool Risks | Mitigated - Credential verification, rate limiting |
| MCP08 | Inadequate Logging | Mitigated - Structured JSON audit logging |
| MCP09 | Lack of Security Documentation | Mitigated - This document |
| MCP10 | Insecure Tool Communication | Mitigated - Stdio transport only (no HTTP exposure) |
### MCP01/MCP07: Credential Validation
The server validates 46elks API credentials on startup before accepting any tool calls:
```
✓ Configuration validated
✓ 46elks credentials verified
✓ MCP SMS Server running on stdio
```
If credentials are invalid, the server will not start:
```
✗ 46elks credential verification failed: 401 Unauthorized
```
### MCP05/MCP07: Rate Limiting
The server implements rate limiting to prevent abuse:
| Tool Category | Default Limit | Environment Variable |
|--------------|---------------|---------------------|
| SMS Sending (`send_sms`) | 10 per minute | `RATE_LIMIT_SMS_PER_MINUTE` |
| Query Tools (all others) | 60 per minute | `RATE_LIMIT_QUERIES_PER_MINUTE` |
When rate limits are exceeded:
```
Error (RATE_LIMIT_EXCEEDED): Rate limit exceeded for SMS sending. Please wait 45 seconds before trying again.
```
### MCP05/MCP06: Input Sanitization
All tool inputs are validated and sanitized:
- **Phone numbers**: International format required (+46...), placeholder numbers blocked
- **Messages**: Control characters stripped, URL detection warnings, length limits
- **Sender IDs**: Impersonation attempts blocked (BANK, POLICE, etc.)
- **Message IDs**: Alphanumeric only, prevents injection attacks
### MCP08: Audit Logging
All tool invocations are logged to stderr in JSON format:
```json
{
"audit": {
"timestamp": "2025-02-05T12:34:56.789Z",
"tool": "send_sms",
"params": {
"to": "+467****123",
"message_length": 42,
"dry_run": false
},
"success": true,
"dryRun": false,
"durationMs": 234
}
}
```
**Privacy protections in audit logs:**
- Phone numbers are masked (+467****123)
- Message content is replaced with length only
- Only metadata (direction, limit, message_id) is logged in full
## Security Best Practices for Users
When using this MCP server:
1. **Keep credentials secure**: Never commit API credentials to version control
2. **Use environment variables**: Store `ELKS_API_USERNAME` and `ELKS_API_PASSWORD` securely
3. **Enable dry run mode**: Always test with `DRY_RUN=true` before sending real SMS
4. **Keep dependencies updated**: Run `npm audit` regularly and update when needed
5. **Use latest version**: Always use the latest release for security patches
6. **Monitor audit logs**: Review stderr output for suspicious activity
7. **Set rate limits**: Adjust `RATE_LIMIT_SMS_PER_MINUTE` for your use case
### Environment Variable Configuration
```bash
# Required credentials
ELKS_API_USERNAME=your-api-username
ELKS_API_PASSWORD=your-api-password
ELKS_PHONE_NUMBER=+46XXXXXXXXX
# Safety settings
DRY_RUN=true # Set to false for production
# Optional rate limiting
RATE_LIMIT_SMS_PER_MINUTE=10
RATE_LIMIT_QUERIES_PER_MINUTE=60
```
## Scope
This security policy covers:
- The 46elks MCP server code in this repository
- Security of the MCP protocol implementation
- Input validation and sanitization
- Audit logging and rate limiting
This policy does **not** cover:
- 46elks API security (report to [46elks support](https://46elks.com/support))
- MCP client security (Claude Desktop, VS Code, etc.)
- Your own API credentials or account security
## Security Features Summary
This MCP server includes the following security features:
- **Credential validation**: API credentials verified before server starts
- **Rate limiting**: Prevents rapid SMS sending (abuse protection)
- **Structured audit logging**: JSON logs for security monitoring
- **Input validation**: Phone numbers, messages, sender IDs validated
- **Control character stripping**: Removes potentially harmful characters
- **URL detection**: Warns about links in SMS content
- **Anti-spoofing**: Blocks impersonation attempts (BANK, POLICE, etc.)
- **Message ID validation**: Prevents injection through malformed IDs
- **Dry run mode**: Prevents accidental SMS sending during development
- **No credential storage**: Credentials passed via environment variables only
- **Local-only operation**: No HTTP endpoints, stdio transport only
- **Dependency scanning**: Automated Dependabot alerts and updates
- **Code scanning**: CodeQL analysis for vulnerability detection
## Disclaimer
This is an unofficial community project and is not affiliated with 46elks AB. For 46elks API security concerns, please contact [46elks support](https://46elks.com/support) directly.
