Feature: Add 'AI Compliance' Strict Configuration Preset

[matthewgoluba]

To ensure data sent to AI models (Gems) adheres to strict compliance policies, we need a pre-defined configuration that enables all relevant patterns by default. This prevents users from manually having to select every single PII type.

Tasks:

• Define a strict-ai preset in packages/core/src/presets.ts (or similar).
• Preset should enable:
  • All PII (Names, Emails, SSN, Phone).
  • All Financial (Credit Cards).
  • All System (UUIDs, IPs, MAC Addresses).
• Expose this preset in the API so clients can request config: strict-ai.

Additional Presets Implemented:

• minimal - Only email and phone
• logs - Format-preserving for IPs, token for secrets
• financial - Focus on credit cards, SSN
• healthcare - HIPAA-focused

[goobz22]

Implementation Notes Based on Codebase Review

Current Configuration System

• packages/core/src/config.ts has ConfigLoader class and DEFAULT_CONFIG
• Config examples exist in config-examples/basic.json and advanced.json
• No preset system exists - only manual JSON configuration

Suggested Implementation

1. Create packages/core/src/presets.ts:

import { RedactorConfig } from './types';

export const PRESETS = {
  'strict-ai': {
    formatOptions: {
      tokenFormat: '[REDACTED_{TYPE}]',
      preserveStructure: false
    },
    patterns: {
      // All PII
      email: { enabled: true, strategy: 'token' },
      phone: { enabled: true, strategy: 'token' },
      ssn: { enabled: true, strategy: 'token' },
      name: { enabled: true, strategy: 'token' },
      // All Financial
      creditCard: { enabled: true, strategy: 'token' },
      creditCardLast4: { enabled: true, strategy: 'token' },
      // All Network/System
      ipv4: { enabled: true, strategy: 'token' },
      ipv6: { enabled: true, strategy: 'token' },
      macAddress: { enabled: true, strategy: 'token' },
      hostname: { enabled: true, strategy: 'token' },
      // System patterns (Issue #2)
      uuid: { enabled: true, strategy: 'token' },
      filePath: { enabled: true, strategy: 'token' },
    }
  },
  
  'pii-only': {
    patterns: {
      email: { enabled: true, strategy: 'mask' },
      phone: { enabled: true, strategy: 'mask' },
      ssn: { enabled: true, strategy: 'mask' },
      name: { enabled: true, strategy: 'mask' },
      creditCard: { enabled: true, strategy: 'mask' },
    }
  },
  
  'logs-safe': {
    // For sanitizing logs before sharing
    patterns: {
      ipv4: { enabled: true, strategy: 'formatPreserving' },
      ipv6: { enabled: true, strategy: 'formatPreserving' },
      macAddress: { enabled: true, strategy: 'formatPreserving' },
      hostname: { enabled: true, strategy: 'formatPreserving' },
      uuid: { enabled: true, strategy: 'formatPreserving' },
      filePath: { enabled: true, strategy: 'token' },
    }
  }
} as const satisfies Record<string, Partial<RedactorConfig>>;

export type PresetName = keyof typeof PRESETS;

export function getPreset(name: PresetName): Partial<RedactorConfig> {
  return PRESETS[name];
}

2. Update engine.ts constructor:

constructor(config?: Partial<RedactorConfig> | string | { preset: PresetName }) {
  if (config && typeof config === 'object' && 'preset' in config) {
    this.config = ConfigLoader.loadFromObject({
      ...getPreset(config.preset),
      ...config // allow overrides
    });
  }
  // ... existing logic
}

3. Export from packages/core/src/index.ts:

export { PRESETS, getPreset, type PresetName } from './presets';

API Integration (Issue #3)

The API endpoint should accept:

{ "text": "...", "preset": "strict-ai" }
// or
{ "text": "...", "config": { "preset": "strict-ai", "patterns": { "name": { "enabled": false } } } }

[goobz22]

Implementation Complete

Created configuration presets in packages/core/src/presets.ts:

Available Presets

• strict-ai - All patterns enabled with token strategy for maximum AI safety
• minimal - Only email and phone patterns
• logs - Format-preserving for IPs, token for secrets (log analysis)
• financial - Focus on credit cards, SSN, account numbers
• healthcare - HIPAA-focused preset

API

import { getPreset, getPresetNames, hasPreset, PRESETS } from '@data-redactor/core'

const config = getPreset('strict-ai')
const names = getPresetNames() // ['strict-ai', 'minimal', 'logs', 'financial', 'healthcare']

Files Changed

• packages/core/src/presets.ts - New file with preset definitions
• packages/core/src/index.ts - Exports

Ready to close.