Quantum Zero-Knowledge Proof (QZKP) Prototype

Author: Nicolas Cloutier (ORCID: 0009-0008-5289-5324)
Organization: GrillCheese AI
Repository: https://github.com/grillcheese-ai/qzkp

What this is

This repository provides an end-to-end reference implementation (prototype) of a quantum zero-knowledge proof (QZKP)-style protocol plus a leakage evaluation test suite designed to catch common implementation failures (e.g., accidental serialization of witness/state-vector data).

Critical note

This repository contains two implementations:

1. QuantumZKP (insecure / educational): demonstrates how transcripts can leak witness structure.
2. SecureQuantumZKP (secure-by-design goals): avoids direct inclusion of witness amplitudes in transcripts and uses cryptographic commitments + PQ authentication.

Important: The “secure” variant is a prototype with explicit assumptions and threat model (see docs/threat_model.md). It is intended for research and review.

Security goals

• Simulator-based computational zero-knowledge claim under stated assumptions.
• Soundness via challenge repetition: soundness error ≤ 2^{-k}.
• Transcript design goal: no direct witness/state-vector inclusion in proofs.
• Post-quantum authentication using ML-DSA (derived from CRYSTALS-Dilithium).

Reproducibility (reviewers)

Local (no IBM access required):

go test -v -run TestReviewerSuite
python reviewer_run_all.py
python make_evidence_manifest.py --write-index



## Quick start

```bash
go test ./... -v
go test ./... -run TestInformationLeakageAnalysis -v
go test ./... -bench=.