Atlas CLI: Machine Learning (ML) Lifecycle & Transparency Manager

⚠️ Disclaimer: This project is currently in active development. The code is not stable and not intended for use in production environments. Interfaces, features, and behaviors are subject to change without notice.

A command-line interface tool for creating, managing, and verifying Content Provenance and Authenticity (C2PA) manifests for machine learning models, datasets, and related artifacts.

Key Features

• Model & Dataset Manifests: Create C2PA-compliant manifests for ML models and datasets
• Cryptographic Signing: Sign manifests with cryptographic keys for authenticity verification, incl. support for the OpenSSF Model Signing (OMS) specification.
• Provenance Linking: Create verifiable links between models, datasets, and ML assets
• Multiple Storage Types: Store manifests in MongoDB, Rekor log, or filesystem backends
• Format Support: Work with models in ONNX, TensorFlow, PyTorch, and Keras formats
• TEE Attestation: Optional support for Trusted Execution Environment (TDX) integration

Installation

Prerequisites

• Rust toolchain (1.70 or later) - Install Rust
• OpenSSL development libraries
• (Optional) Protobuf compiler for TDX support

Install Methods

Install from crates.io

The simplest way to install Atlas CLI is using cargo:

cargo install atlas-cli

Install with Specific Features

With TDX Attestation Support:

# First install protobuf compiler
# Ubuntu/Debian:
sudo apt install protobuf-compiler

# Then install with TDX feature
cargo install atlas-cli --features with-tdx

Install from Source

# Clone repositories
git clone https://github.com/IntelLabs/atlas-cli
cd atlas-cli

# Build and install
cargo install --path .

# Or build without installing
cargo build --release
# Binary will be at ./target/release/atlas-cli

# To update to the latest version:
cargo install atlas-cli --force

Documentation

For more detailed information, please refer to:

• User Guide - Installation, configuration, and command reference
• Development Guide - Contributing, building, and architecture
• Examples - Usage examples and workflow patterns

License

This project is licensed under the Apache 2.0 License - see the LICENSE file for details.

Citation

If you use Atlas CLI in your research or work, please cite our paper:

@misc{atlas2025github,
      title={Atlas: A Framework for ML Lifecycle Provenance & Transparency},
      author={Marcin Spoczynski and Marcela S. Melara and Sebastian Szyller},
      year={2025},
      eprint={2502.19567},
      archivePrefix={arXiv},
      primaryClass={cs.CR},
      url={https://arxiv.org/abs/2502.19567v1}
}

Related Resources

• Paper: Atlas: A Framework for ML Lifecycle Provenance & Transparency
• Blog Post: Building Trust in AI: An End-to-End Approach for the Machine Learning Lifecycle
• Documentation: [docs.rs/atlas-cli]
• Crate: [crates.io/crates/atlas-cli]