Padding Oracle on Aws

The correct mitigation is to remove the oracle entirely: ensure that decrypt operations are never exposed to unauthenticated callers and that error responses do not leak padding‑validation details. Use AWS native controls as follows:

1. Authenticate and authorize every decrypt request with least‑privilege IAM policies.
2. Use envelope encryption: encrypt data with a data key, then encrypt the data key with KMS. Store only the encrypted data key; never expose a raw decrypt API.
3. Prefer authenticated encryption modes (e.g., AES‑GCM) which provide integrity and eliminate padding‑oracle vectors. If you must use CBC mode, always compute an HMAC over the ciphertext before decryption and verify it first.
4. Normalize error responses: catch all KMS exceptions and return a generic failure (HTTP 500) with no indication of why decryption failed.
5. Leverage API Gateway request validation and AWS WAF to block requests with malformed ciphertexts or abnormal frequencies.
6. When possible, avoid custom decrypt endpoints altogether and rely on AWS Secrets Manager or Parameter Store with SecureString values, which return plaintext only to callers with the appropriate kms:Decrypt permission.

Revised Node.js example using proper authorization and generic error handling:

const { KMSClient, DecryptCommand } = require('@aws-sdk/client-kms');
const kms = new KMSClient({});
exports.handler = async (event) => {
  // Verify caller identity via IAM role / JWT before proceeding
  const auth = event.requestContext.authorizer?.iam;
  if (!auth) { return { statusCode: 403, body: 'Forbidden' }; }

  try {
    const ciphertext = Buffer.from(event.body, 'base64');
    const data = await kms.send(new DecryptCommand({ CiphertextBlob: ciphertext }));
    return { statusCode: 200, body: data.Plaintext.toString('base64') };
  } catch (err) {
    // Generic error – no padding‑oracle leakage
    return { statusCode: 500, body: 'Decryption failed' };
  }
};

Equivalent Python (boto3) snippet:

import boto3, json, os
kms = boto3.client('kms')
def lambda_handler(event, context):
    # Simple IAM check – adjust to your auth method
    if not event.get('requestContext', {}).get('authorizer', {}).get('iam'):
        return {'statusCode': 403, 'body': json.dumps('Forbidden')}
    try:
        blob = bytes.fromhex(event['body'])
        resp = kms.decrypt(CiphertextBlob=blob)
        return {'statusCode': 200, 'body': resp['Plaintext'].decode()}
    except Exception:
        return {'statusCode': 500, 'body': json.dumps('Decryption failed')}

After applying these controls, rescan with middleBrick to confirm the finding is resolved.