- Bcrypt hashing for API keys (SHA-256 legacy support)
- Rate limiting per workspace with burst allowance
- HMAC-SHA256 webhook signing
- Scope-based permissions
### SQL Injection Prevention
- Multi-layer query validation (MCP package)
- Blocked keywords (INSERT, UPDATE, DELETE, DROP)
- Pattern detection for SQL injection attempts
- Read-only database connection support
- Table access controls
### Input Sanitization
- Built-in HTML/JS sanitization
- XSS prevention
- Email validation and disposable email blocking
### Security Headers
- Content Security Policy (CSP)
- HSTS, X-Frame-Options, X-Content-Type-Options
- Referrer Policy
- Permissions Policy
### Action Gate System
- Request whitelisting for sensitive operations
- Training mode for development
- Audit logging for all actions
## Security Best Practices
When using the Core PHP Framework:
### API Keys
- Store API keys securely (never in version control)
- Use environment variables or secure key management
- Rotate keys regularly
- Use minimal required scopes
### Database Access
- Use read-only connections for MCP tools
- Configure blocked tables for sensitive data
- Enable query whitelisting in production
### Workspace Context
- Always validate workspace context in custom tools
- Use `RequiresWorkspaceContext` trait
- Never bypass workspace scoping
### Rate Limiting
- Configure appropriate limits per tier
- Monitor rate limit violations
- Implement backoff strategies in API clients
### Activity Logging
- Enable activity logging for sensitive operations
- Regularly review activity logs
- Set appropriate retention periods
## Security Changelog
See [packages/core-mcp/changelog/2026/jan/security.md](packages/core-mcp/changelog/2026/jan/security.md) for recent security fixes.
## Credits
We appreciate the security research community and would like to thank the following researchers for responsibly disclosing vulnerabilities:
- *No vulnerabilities reported yet*
## Bug Bounty Program
We do not currently have a formal bug bounty program, but we deeply appreciate security research. Researchers who report valid security vulnerabilities will be:
- Credited in our security advisories (if desired)
- Listed in this document
- Given early access to security patches
## PGP Key
For sensitive security reports, you may encrypt your message using our PGP key: