pycryptox 3.0.1

Post-quantum cryptography library with built-in plausible deniability.

Pycryptox

Post-quantum cryptography, plausible deniability, secure software, secure hardware, and more.

Pycryptox is a Python library that takes cryptographic engineering seriously. Every primitive choice, every wire format, every API decision is documented and motivated against a concrete threat model. The goal is not "encrypt this for me" — it is "protect this against the adversary I am actually worried about, today and after a quantum computer arrives."

import pycryptox as crx

keys = crx.blue.genkeys("2")
bundle = crx.blue.encrypt("2", keys["gpubkey"], xmsg="real message", ymsg="credible decoy")

# Same bundle, two recipients, two different plaintexts:
crx.blue.decrypt("2", keys["xprivkey"], bundle, slot="x")  # → "real message"
crx.blue.decrypt("2", keys["yprivkey"], bundle, slot="y")  # → "credible decoy"

# Under coercion, the user reveals only yprivkey. The adversary obtains the
# decoy, has no cryptographic way to prove the real message exists, and the
# real key is physically elsewhere on a separate device. This is BLUE.

What is in the box

Protocol	What it does	Primary use
PURPLE	Password-based encryption with adjustable Argon2id strength	Brain-only secrets you can recover from a passphrase alone
BLUE	Deniable asymmetric encryption with plausible-deniability dual-channel	Sensitive content where coercion is a realistic threat
RED	Threshold encryption (t-of-n secret sharing)	Distributed trust, recovery shares, multi-party access
YELLOW	Steganography (placeholder, coming soon)	Concealing the existence of a message itself
Keyx	Self-contained key storage with two trust tiers (keys / ckeys)	On-device persistence with physical separation of critical keys

All asymmetric work is post-quantum from day one: ML-KEM-512 (FIPS 203) via liboqs handles every key encapsulation, and ChaCha20-Poly1305 handles every symmetric step. No RSA, no ECC, no classical Diffie-Hellman.

Why a serious project

This is not a hobby toy. Pycryptox was designed for cases where getting the threat model wrong has real consequences: journalists protecting sources, activists under surveillance, custody of long-term secrets that must outlive today's cryptanalytic state. The library is honest about what it does and does not protect against — every protocol has a dedicated Me vs Opponent section in its documentation that walks through the attacks a competent adversary will try, the design decision that defeats each one, and the attacks that are explicitly out of scope.

The codebase reflects that posture:

• Versioning is mandatory: every bundle carries its protocol version, every migration is explicit through update_bundle.
• Old versions are frozen byte-for-byte after release. Cryptanalysis of a deployed version cannot be silently patched away.
• Bundle structures are exhaustively documented down to the byte.
• 264+ tests covering all protocols, Keyx storage, version dispatch, and error paths.

Feedback I am looking for

This library is maintained by a single person under the Cryptox identity, and I want serious feedback to improve it: cryptanalysis attempts on the protocols, threat-model holes, API criticism, documentation clarity, performance issues on real workloads. Send anything substantive to:

📧 [e-mail]

I read everything. I will not always agree, but I will engage. The library is at its best when it is challenged.

Proposing features

If you have a use case Pycryptox does not yet cover, or a protocol idea that fits the project's posture (post-quantum, deniability-aware, explicit threat model), I am open to discussion. Same address: [e-mail]. Please describe:

1. The threat model (what is the adversary, what are their capabilities, what do they want)
2. The use case (who needs this, in what context)
3. The deniability / security properties you would expect

I will not implement features without a clear motivation, but a good motivation goes a long way.

Prerequisites

Pycryptox uses liboqs (via liboqs-python) for all ML-KEM-512 post-quantum operations. This dependency is not installed automatically because the current PyPI release of liboqs-python is broken — it must be installed from GitHub:

pip install git+https://github.com/open-quantum-safe/liboqs-python@0.12.0

This requires a C compiler, CMake, and Git. The first import oqs automatically downloads and compiles the liboqs C library (~2-5 min, one-time only). See the installation guide for platform-specific instructions.

Installation

pip install pycryptox

Documentation

📖 English · Français

Each protocol has its own document with cryptographic primitives, wire format, threat model (Me vs Opponent), API reference, complete examples, and version history.

Tests

🧪 Test suite — 269 tests covering all protocols and Keyx.

cd pycryptox/tests
python run_all.py

Acknowledgments

📜 See THANKS.md — Pycryptox builds on the work of Open Quantum Safe, PyCA, the Argon2 team, the EFF, and many others.

Credits

See CREDITS.md

License

MIT — see LICENSE.