Implement security best practices for Lindy AI integrations. Use when securing API keys, configuring permissions, or implementing security controls. Trigger with phrases like "lindy security", "secure lindy", "lindy API key security", "lindy permissions".
Essential security practices for Lindy AI integrations.
// NEVER do this
const apiKey = 'lnd_abc123...'; // Hardcoded - BAD!
// DO this instead
const apiKey = process.env.LINDY_API_KEY;
// Or use secret management
import { SecretManager } from '@google-cloud/secret-manager';
async function getApiKey(): Promise<string> {
const client = new SecretManager();
const [secret] = await client.accessSecretVersion({
name: 'projects/my-project/secrets/lindy-api-key/versions/latest',
});
return secret.payload?.data?.toString() || '';
}
# .env.development
LINDY_API_KEY=lnd_dev_xxx
LINDY_ENVIRONMENT=development
# .env.production
LINDY_API_KEY=lnd_prod_xxx
LINDY_ENVIRONMENT=production
// Validate environment
function validateEnvironment(): void {
const env = process.env.LINDY_ENVIRONMENT;
const key = process.env.LINDY_API_KEY;
if (!key) {
throw new Error('LINDY_API_KEY not set');
}
if (env === 'production' && key.startsWith('lnd_dev_')) {
throw new Error('Development key used in production!');
}
}
import { Lindy } from '@lindy-ai/sdk';
const lindy = new Lindy({ apiKey: process.env.LINDY_API_KEY });
async function createSecureAgent() {
const agent = await lindy.agents.create({
name: 'Secure Agent',
instructions: 'Handle data securely.',
permissions: {
// Restrict to specific tools
allowedTools: ['email', 'calendar'],
// Prevent external network access
networkAccess: 'internal-only',
// Limit data access
dataScopes: ['read:users', 'write:tickets'],
},
});
return agent;
}
async function withAuditLog<T>(
operation: string,
fn: () => Promise<T>
): Promise<T> {
const start = Date.now();
const requestId = crypto.randomUUID();
console.log(JSON.stringify({
type: 'audit',
operation,
requestId,
timestamp: new Date().toISOString(),
status: 'started',
}));
try {
const result = await fn();
console.log(JSON.stringify({
type: 'audit',
operation,
requestId,
duration: Date.now() - start,
status: 'completed',
}));
return result;
} catch (error: any) {
console.log(JSON.stringify({
type: 'audit',
operation,
requestId,
duration: Date.now() - start,
status: 'failed',
error: error.message,
}));
throw error;
}
}
[ ] API keys stored in environment variables or secret manager
[ ] Different keys for dev/staging/prod environments
[ ] Key validation on startup
[ ] Agent permissions configured (least privilege)
[ ] Audit logging enabled
[ ] Network access restricted where possible
[ ] Regular key rotation scheduled
[ ] Access reviewed quarterly
| Risk | Mitigation | Implementation |
|---|---|---|
| Key exposure | Secret manager | Use cloud secrets |
| Wrong env | Validation | Check key prefix |
| Over-permission | Least privilege | Restrict agent tools |
| No audit | Logging | Log all operations |
// security/index.ts
export async function initializeLindy(): Promise<Lindy> {
// Validate environment
validateEnvironment();
// Get key from secret manager
const apiKey = await getApiKey();
// Initialize with security options
const lindy = new Lindy({
apiKey,
timeout: 30000,
retries: 3,
});
// Verify connection
await lindy.users.me();
console.log('Lindy initialized securely');
return lindy;
}
Proceed to lindy-prod-checklist for production readiness.