escrowai-encrypt 0.0.2

CLI tool and Python library for encrypting algorithms and datasets with AES-256-GCM for secure computation on EscrowAI

escrowai-encrypt

A Python package and CLI tool for encrypting algorithms and datasets for use with EscrowAI. Implements secure encryption practices using AES-256-GCM for data encryption and RSA-OAEP for key wrapping.

Overview

The escrowai-encrypt package provides:

• Command-line tool for encrypting/decrypting files and datasets
• Python library for integrating encryption into your workflows
• Azure Blob Storage integration for secure cloud data handling
• Algorithm protection with selective file encryption and secrets management

Installation

pip install escrowai-encrypt

This installs both the escrowai-encrypt CLI tool and the Python library.

Quick Start (CLI)

1. Generate an Encryption Key

escrowai-encrypt generate-key --key-output my.key

2. Encrypt an Algorithm

Encrypt your algorithm code while excluding configuration files:

escrowai-encrypt encrypt-algorithm \
  --input ./my-algorithm \
  --key my.key \
  --output ./encrypted-algorithm \
  --exclude Dockerfile requirements.txt run.sh

This creates an encrypted version of your algorithm with a secrets.yaml manifest.

3. Create a Packaged Algorithm

Encrypt and package your algorithm as a zip file:

escrowai-encrypt encrypt-algorithm \
  --input ./my-algorithm \
  --key my.key \
  --exclude Dockerfile requirements.txt \
  --zip algorithm-encrypted.zip

4. Encrypt and Upload Dataset to Azure

escrowai-encrypt encrypt \
  --input ./my-dataset \
  --key my.key \
  --sas-url "https://storage.blob.core.windows.net/container?sv=..."

5. Decrypt Dataset from Azure

escrowai-encrypt decrypt \
  --sas-url "https://storage.blob.core.windows.net/container?sv=..." \
  --key my.key \
  --output ./decrypted-data

CLI Reference

Commands

generate-key - Generate a new Content Encryption Key (CEK)

escrowai-encrypt generate-key [--key-output FILENAME]

encrypt-algorithm - Encrypt an algorithm directory with selective exclusions

escrowai-encrypt encrypt-algorithm \
  --input INPUT_FOLDER \
  --key KEY_FILE \
  --output OUTPUT_FOLDER \
  [--exclude FILE1 FILE2 ...] \
  [--zip OUTPUT.zip] \
  [--debug]

encrypt - Encrypt and upload dataset to Azure Blob Storage

escrowai-encrypt encrypt \
  --input LOCAL_FOLDER \
  --key KEY_FILE \
  --sas-url AZURE_SAS_URL \
  [--debug]

# Or encrypt from one blob storage to another
escrowai-encrypt encrypt \
  --source-sas-url SOURCE_AZURE_SAS_URL \
  --key KEY_FILE \
  --sas-url TARGET_AZURE_SAS_URL

decrypt - Decrypt and download dataset from Azure Blob Storage

escrowai-encrypt decrypt \
  --sas-url AZURE_SAS_URL \
  --key KEY_FILE \
  --output LOCAL_FOLDER \
  [--debug]

Options

• --input - Input file or folder path
• --key - Content Encryption Key file path
• --output - Output path for decrypted files or encrypted algorithms
• --sas-url - Azure Blob Storage SAS URL
• --source-sas-url - Source Azure Blob Storage SAS URL (for blob-to-blob encryption)
• --key-output - Output filename for generated key
• --exclude - List of file names to exclude from algorithm encryption (e.g., Dockerfile, requirements.txt)
• --zip - Create a zip file of the encrypted algorithm
• --debug - Enable debug output

Python API

For advanced use cases, you can use the Python library directly in your code.

Key Management

from escrowai_encrypt.encryption import generate_content_encryption_key

# Generate a new encryption key
generate_content_encryption_key('my_key.key')

Algorithm Encryption

from escrowai_encrypt.encryption import encrypt_algorithm

# Encrypt an algorithm directory
encrypt_algorithm(
    algorithm_directory='path/to/algorithm',
    content_encryption_key='my_key.key',
    filename='encrypted_algorithm.zip'
)

Dataset Encryption

from escrowai_encrypt.encryption import encrypt_upload_dataset

# Encrypt and upload a dataset to Azure Blob Storage
encrypt_upload_dataset(
    dataset_directory='path/to/dataset',
    content_encryption_key='my_key.key',
    dataset_sas_uri='https://storage-account.blob.core.windows.net/container?sv=...'
)

Blob-to-Blob Encryption

from escrowai_encrypt.encryption import encrypt_upload_dataset_from_blob

# Encrypt data from one blob storage to another
encrypt_upload_dataset_from_blob(
    dataset_sas_uri_unencrypted='https://source.blob.core.windows.net/container?sv=...',
    content_encryption_key='my_key.key',
    dataset_sas_uri='https://target.blob.core.windows.net/container?sv=...'
)

Decryption

from escrowai_encrypt.decryption import decrypt_secret

# Decrypt an encrypted file
decrypt_secret(
    secret='encrypted_file.bkenc',
    content_encryption_key='my_key.key',
    filename='decrypted_file.txt'
)

Key Wrapping

from escrowai_encrypt.encryption import generate_wrapped_content_encryption_key

# Wrap a CEK with a Key Encryption Key (KEK)
generate_wrapped_content_encryption_key(
    content_encryption_key='my_key.key',
    key_encryption_key='public_key.pem',
    filename='wrapped_key.bkenc'
)

Examples

See the examples/ folder for more comprehensive usage examples, including a full-featured CLI implementation that demonstrates advanced patterns.

Security Features

• AES-256-GCM encryption for all data
• PBKDF2 key derivation with 10,000 iterations and random salts
• RSA-OAEP key wrapping for secure key management
• 16MB chunk processing for efficient handling of large files
• Salted encryption - All encrypted files include a random 8-byte salt

Algorithm Encryption Features

When encrypting algorithms with encrypt-algorithm:

• Selective exclusion - Exclude files like Dockerfile, requirements.txt, or run.sh
• Automatic Dockerfile detection - Dockerfiles are automatically excluded
• secrets.yaml generation - Creates a manifest mapping encrypted files to originals
• Optional packaging - Use --zip to create a ready-to-deploy package

Requirements

• Python >= 3.6
• Dependencies (automatically installed):
  • azure-storage-blob - Azure Blob Storage integration
  • cryptography - Encryption primitives
  • pyyaml - YAML file generation

License

MIT License - Copyright (c) 2024 BeeKeeperAI, Inc.

Support

For issues and questions, please visit the GitHub repository.