# Security Policy
## Supported Versions
Currently supported versions with security updates:
| Version | Supported |
| ------- | ------------------ |
| 1.0.x | :white_check_mark: |
## Reporting a Vulnerability
**Please DO NOT report security vulnerabilities through public GitHub issues.**
Instead, please report security vulnerabilities by emailing the maintainers with the subject line "SECURITY VULNERABILITY - Airtable MCP Server".
Include the following information:
- Type of vulnerability
- Full paths of affected source file(s)
- Location of the affected code (tag/branch/commit/direct URL)
- Step-by-step instructions to reproduce the issue
- Proof-of-concept or exploit code (if possible)
- Impact of the vulnerability
### What to expect
- **Acknowledgment**: You will receive an acknowledgment within 48 hours
- **Updates**: We will send regular updates about our progress
- **Disclosure**: We follow coordinated disclosure practices
- **Credit**: If you wish, we will credit you in the security advisory
## Security Features
### Input Validation
All tool inputs are validated using Zod schemas with:
- Maximum string lengths (1000 characters)
- Maximum array sizes (100 items for most arrays)
- Maximum record limits (1000 records per query)
- Field name sanitization (alphanumeric + safe special characters only)
- Table name sanitization
- Record ID format validation (`recXXXXXXXXXXXXXX`)
### Formula Injection Protection
- **Escaping**: All user-provided values in formulas are escaped using SQL-style escaping (single quote → double single quote)
- **Field Name Validation**: Field names are sanitized to prevent breaking out of `{FieldName}` context
- **Number Validation**: Numeric values validated as finite numbers to prevent injection via NaN/Infinity
- **Date Validation**: Date values validated against ISO 8601 format
### Error Handling
- **Sanitized Errors**: Error messages are sanitized before being returned to clients
- **No Stack Traces**: Stack traces never exposed to clients (logged server-side only)
- **Credential Redaction**: API keys, base IDs, and tokens automatically redacted from error messages
- **Custom Error Classes**: Type-safe error handling with appropriate HTTP status codes
### Data Protection
- **Attachment Filtering**: Automatically excludes large attachment fields by default to prevent data leaks
- **Field Exclusion**: Supports excluding sensitive fields from responses
- **Long Text Filtering**: Can exclude potentially sensitive long text fields
- **Preset Filters**: Pre-configured field sets (minimal, contact, summary) for safe queries
### Authentication & Authorization
- **PAT Validation**: Personal Access Token validated at startup
- **PAT-Only Authentication**: Only supports Personal Access Tokens (legacy API keys deprecated)
- **No Token Storage**: PATs only stored in environment variables, never in code
- **Token Rotation**: Supports seamless token rotation via environment variable updates
- **Scoped Permissions**: Leverages Airtable's PAT permission system for fine-grained access control
## Security Best Practices
### Environment Variables
**CRITICAL**: Never commit `.env` files to version control!
```bash
# .env (DO NOT COMMIT)
AIRTABLE_PAT=patXXXXXXXXXXXXXXXXXX.XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
AIRTABLE_BASE_ID=appXXXXXXXXXXXXXX
```
Add to `.gitignore`:
```gitignore
.env
.env.local
.env.*.local
```
### Personal Access Token (PAT) Security
1. **Use Scoped Permissions**: Create PATs with minimum required scopes (e.g., read-only if possible)
2. **Rotate Regularly**: Rotate PATs every 90 days
3. **Separate Environments**: Use different PATs for development/staging/production
4. **Monitor Usage**: Regularly check Airtable API usage for anomalies
5. **Revoke Unused Tokens**: Remove PATs that are no longer in use
6. **Name Tokens Clearly**: Use descriptive names to track where each PAT is used
### Deployment Security
#### Claude Desktop Configuration
```json
{
"mcpServers": {
"airtable": {
"command": "node",
"args": ["/absolute/path/to/build/index.js"],
"env": {
"AIRTABLE_PAT": "patXXXXXXXXXXXXXXXXXX.XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX",
"AIRTABLE_BASE_ID": "appXXXXXXXXXXXXXX"
}
}
}
}
```
**Security Notes**:
- Use absolute paths to prevent path traversal
- Store config file with restricted permissions: `chmod 600 claude_desktop_config.json`
- Never commit config files with real PATs
- Use separate PATs for each MCP server instance
- Limit PAT scopes to only what's needed (principle of least privilege)
#### Production Deployment
1. **Use Secrets Management**:
- AWS Secrets Manager
- HashiCorp Vault
- Kubernetes Secrets
- Azure Key Vault
2. **Principle of Least Privilege**:
- Run MCP server with minimal user permissions
- Use read-only API keys when possible
- Limit base access to required tables only
3. **Network Security**:
- Run behind firewall if exposing over network
- Use TLS/SSL for all network communication
- Implement IP whitelisting if applicable
4. **Monitoring & Logging**:
- Enable structured logging
- Monitor for unusual API patterns
- Set up alerts for authentication failures
- Track API rate limit usage
### Field Filtering (IMPORTANT for Claude Desktop)
**Always use `excludeAttachments=true`** to prevent timeouts and data leakage:
```json
{
"table": "Contacts",
"excludeAttachments": true, // Prevents loading images/files
"excludeLongText": true, // Prevents loading large text fields
"maxRecords": 100 // Limit response size
}
```
Or use presets for common scenarios:
```json
{
"table": "Contacts",
"preset": "contact", // Only returns contact info (no attachments)
"maxRecords": 50
}
```
Available presets:
- `minimal`: Name and email only
- `contact`: Contact information only
- `summary`: Key fields without attachments
- `full`: All fields (use with caution!)
### Rate Limiting
Airtable has rate limits:
- **5 requests per second per base**
- **100,000 API calls per workspace per day**
To avoid hitting limits:
1. Use `maxRecords` to limit response size
2. Implement client-side caching where appropriate
3. Batch operations when possible
4. Monitor API usage in Airtable dashboard
## Known Security Considerations
### Formula Execution
Custom formulas in `airtable_smart_query` can execute Airtable formula functions. While these are sandboxed by Airtable, be cautious with user-provided formula strings.
**Recommendation**: Prefer using the built-in query builders (`ageRange`, `multipleSelect`, `numberRange`) over `customFormula` when possible.
### Claude Desktop Integration
When using with Claude Desktop:
- Claude has access to ALL tools defined in this MCP server
- Claude can read/write ANY data in the connected Airtable base
- Consider using separate bases for sensitive vs. non-sensitive data
- Use Airtable permissions to restrict what the API key can access
### Read-Only Fields
The following field types are read-only and attempting to set them will throw errors:
- `formula`, `rollup`, `count`, `lookup`
- `createdTime`, `createdBy`, `lastModifiedTime`, `lastModifiedBy`
- `autoNumber`
- `aiText`, `aiImage` (AI-generated fields)
This is a security feature to prevent unintended data modification.
## Security Checklist for Production
- [ ] API keys stored in environment variables (not code)
- [ ] `.env` files added to `.gitignore`
- [ ] Using scoped API keys with minimum permissions
- [ ] API key rotation schedule established
- [ ] Structured logging configured
- [ ] Error monitoring/alerting configured
- [ ] Rate limit monitoring configured
- [ ] Regular security updates scheduled
- [ ] Access logs reviewed regularly
- [ ] Using `excludeAttachments=true` by default
- [ ] Field filtering configured appropriately
- [ ] Running with minimal user permissions
- [ ] Dependencies audit (`npm audit`) passing
- [ ] HTTPS enforced for all attachments
- [ ] Deployment behind firewall/VPN if applicable
## Updates and Patches
Subscribe to security advisories:
- Watch this repository for security updates
- Enable GitHub Dependabot alerts
- Run `npm audit` regularly
- Keep dependencies up to date
## Additional Resources
- [Airtable API Security](https://airtable.com/developers/web/api/authentication)
- [OWASP API Security Top 10](https://owasp.org/www-project-api-security/)
- [Model Context Protocol Security](https://modelcontextprotocol.io/docs/security)
## Contact
For security concerns, please contact the maintainers directly rather than opening public issues.
---
**Last Updated**: 2025-01-23
**Version**: 1.0.0
