@helixdevelopment/lindy-data-handling

---
name: lindy-data-handling
description: |
  Best practices for handling data with Lindy AI.
  Use when managing sensitive data, implementing data privacy,
  or ensuring data compliance.
  Trigger with phrases like "lindy data", "lindy privacy",
  "lindy PII", "lindy data handling", "lindy GDPR".
allowed-tools: Read, Write, Edit
version: 1.0.0
license: MIT
author: Jeremy Longshore <jeremy@intentsolutions.io>
---

# Lindy Data Handling

## Overview
Best practices for secure and compliant data handling with Lindy AI.

## Prerequisites
- Understanding of data privacy requirements
- Knowledge of applicable regulations (GDPR, CCPA, HIPAA)
- Access to data classification documentation

## Instructions

### Step 1: Data Classification
```typescript
// data/classification.ts
enum DataClassification {
  PUBLIC = 'public',
  INTERNAL = 'internal',
  CONFIDENTIAL = 'confidential',
  RESTRICTED = 'restricted', // PII, PHI, etc.
}

interface DataPolicy {
  classification: DataClassification;
  canSendToLindy: boolean;
  requiresRedaction: boolean;
  retentionDays: number;
}

const policies: Record<DataClassification, DataPolicy> = {
  [DataClassification.PUBLIC]: {
    classification: DataClassification.PUBLIC,
    canSendToLindy: true,
    requiresRedaction: false,
    retentionDays: 365,
  },
  [DataClassification.INTERNAL]: {
    classification: DataClassification.INTERNAL,
    canSendToLindy: true,
    requiresRedaction: false,
    retentionDays: 90,
  },
  [DataClassification.CONFIDENTIAL]: {
    classification: DataClassification.CONFIDENTIAL,
    canSendToLindy: true,
    requiresRedaction: true,
    retentionDays: 30,
  },
  [DataClassification.RESTRICTED]: {
    classification: DataClassification.RESTRICTED,
    canSendToLindy: false,
    requiresRedaction: true,
    retentionDays: 7,
  },
};
```

### Step 2: PII Detection and Redaction
```typescript
// data/pii-redactor.ts
interface PIIPattern {
  name: string;
  pattern: RegExp;
  replacement: string;
}

const piiPatterns: PIIPattern[] = [
  {
    name: 'email',
    pattern: /[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}/g,
    replacement: '[EMAIL REDACTED]',
  },
  {
    name: 'phone',
    pattern: /(\+\d{1,3}[-.]?)?\(?\d{3}\)?[-.]?\d{3}[-.]?\d{4}/g,
    replacement: '[PHONE REDACTED]',
  },
  {
    name: 'ssn',
    pattern: /\d{3}-\d{2}-\d{4}/g,
    replacement: '[SSN REDACTED]',
  },
  {
    name: 'credit_card',
    pattern: /\d{4}[-\s]?\d{4}[-\s]?\d{4}[-\s]?\d{4}/g,
    replacement: '[CREDIT CARD REDACTED]',
  },
  {
    name: 'ip_address',
    pattern: /\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}/g,
    replacement: '[IP REDACTED]',
  },
];

export function redactPII(text: string): { redacted: string; found: string[] } {
  let redacted = text;
  const found: string[] = [];

  for (const pattern of piiPatterns) {
    const matches = text.match(pattern.pattern);
    if (matches) {
      found.push(pattern.name);
      redacted = redacted.replace(pattern.pattern, pattern.replacement);
    }
  }

  return { redacted, found };
}
```

### Step 3: Secure Data Pipeline
```typescript
// lib/secure-lindy.ts
import { Lindy } from '@lindy-ai/sdk';
import { redactPII } from '../data/pii-redactor';

export class SecureLindy {
  private lindy: Lindy;

  constructor() {
    this.lindy = new Lindy({ apiKey: process.env.LINDY_API_KEY });
  }

  async runSecure(agentId: string, input: string, options: {
    classification: DataClassification;
    redactPII?: boolean;
  }) {
    // Check if data can be sent to Lindy
    const policy = policies[options.classification];
    if (!policy.canSendToLindy) {
      throw new Error(`Data classification ${options.classification} cannot be sent to Lindy`);
    }

    // Redact PII if required
    let processedInput = input;
    let redactionLog: string[] = [];

    if (policy.requiresRedaction || options.redactPII) {
      const result = redactPII(input);
      processedInput = result.redacted;
      redactionLog = result.found;

      if (redactionLog.length > 0) {
        console.log('PII redacted:', redactionLog);
      }
    }

    // Run agent
    const result = await this.lindy.agents.run(agentId, { input: processedInput });

    // Log for compliance
    await this.logDataAccess({
      agentId,
      classification: options.classification,
      piiRedacted: redactionLog,
      timestamp: new Date(),
    });

    return result;
  }

  private async logDataAccess(log: any): Promise<void> {
    // Store audit log
    console.log('Data access log:', JSON.stringify(log));
  }
}
```

### Step 4: Data Retention Management
```typescript
// data/retention.ts
import { Lindy } from '@lindy-ai/sdk';

interface RetentionPolicy {
  maxAgeDays: number;
  autoDelete: boolean;
}

async function enforceRetention(policy: RetentionPolicy) {
  const lindy = new Lindy({ apiKey: process.env.LINDY_API_KEY });

  const cutoffDate = new Date();
  cutoffDate.setDate(cutoffDate.getDate() - policy.maxAgeDays);

  // Get old runs
  const runs = await lindy.runs.list({
    before: cutoffDate.toISOString(),
  });

  console.log(`Found ${runs.length} runs older than ${policy.maxAgeDays} days`);

  if (policy.autoDelete) {
    for (const run of runs) {
      await lindy.runs.delete(run.id);
      console.log(`Deleted run: ${run.id}`);
    }
  }
}
```

### Step 5: GDPR Compliance
```typescript
// compliance/gdpr.ts
import { Lindy } from '@lindy-ai/sdk';

class GDPRHandler {
  private lindy: Lindy;

  constructor() {
    this.lindy = new Lindy({ apiKey: process.env.LINDY_API_KEY });
  }

  // Right to Access
  async exportUserData(userId: string): Promise<any> {
    const runs = await this.lindy.runs.list({ userId });
    const agents = await this.lindy.agents.list({ userId });

    return {
      exportDate: new Date().toISOString(),
      userId,
      runs: runs.map(r => ({
        id: r.id,
        agentId: r.agentId,
        createdAt: r.createdAt,
        // Don't include input/output for security
      })),
      agents: agents.map(a => ({
        id: a.id,
        name: a.name,
        createdAt: a.createdAt,
      })),
    };
  }

  // Right to Erasure
  async deleteUserData(userId: string): Promise<void> {
    // Delete all runs
    const runs = await this.lindy.runs.list({ userId });
    for (const run of runs) {
      await this.lindy.runs.delete(run.id);
    }

    // Delete all agents
    const agents = await this.lindy.agents.list({ userId });
    for (const agent of agents) {
      await this.lindy.agents.delete(agent.id);
    }

    console.log(`Deleted all data for user: ${userId}`);
  }
}
```

## Data Handling Checklist
```markdown
[ ] Data classification scheme defined
[ ] PII detection implemented
[ ] Redaction applied before sending to Lindy
[ ] Audit logging enabled
[ ] Retention policies defined
[ ] GDPR/CCPA handlers implemented
[ ] Data access controls configured
[ ] Encryption in transit (HTTPS)
[ ] Regular data audits scheduled
```

## Output
- Data classification system
- PII detection and redaction
- Secure data pipeline
- Retention management
- GDPR compliance handlers

## Error Handling
| Issue | Cause | Solution |
|-------|-------|----------|
| PII leaked | Missing redaction | Enable auto-redaction |
| Retention exceeded | No cleanup | Schedule retention job |
| Classification missing | No policy | Default to restricted |

## Resources
- [Lindy Privacy Policy](https://lindy.ai/privacy)
- [Lindy Security](https://lindy.ai/security)
- [GDPR Guidelines](https://gdpr.eu/)

## Next Steps
Proceed to `lindy-enterprise-rbac` for access control.