Q-Day Time Machine

Find the cryptography a quantum computer will break, see when it breaks, prove it on real IBM quantum hardware, and get the post-quantum fix.

Try it live Get started Watch the demo View on GitHub

pip install qday-time-machine
qday scan --path ./src --fail-on critical

45 / 45validation tests passed on three IBM quantum computers

~3×10⁻³error per two-qubit gate we measured on the best qubits

~4 Mphysical qubits RSA-2048 would need at that error rate

1,002automated tests, 96% line coverage

The problem

Shor's algorithm will break the RSA and elliptic-curve cryptography behind today's logins, payments and signatures. Attackers can record encrypted traffic now and decrypt it later ("harvest now, decrypt later"), and NIST's draft transition plan retires RSA and elliptic-curve cryptography by 2035. Teams know the threat exists, but can't say which systems break first, in what year, or whether the data they send today is already at risk.

What it does

Step	What you get
1. Scan	Every use of cryptography in a website's TLS setup or a code base (eight languages), including detection of the hybrid post-quantum group X25519MLKEM768.
2. Inventory	A CycloneDX 1.6 cryptography bill of materials (CBOM).
3. Time Machine	A break year for each algorithm: published attack costs (logical qubits and gates) against IBM's quantum roadmap.
4. Risk	Mosca's inequality and a harvest-now window: what to fix first, and which recorded data is already exposed.
5. Prove it	Shor's and Grover's algorithms in Qiskit, run on ibm_fez, ibm_kingston and ibm_marrakesh with pre-registered pass rules, and stepped through gate by gate with quantum-debugger.
6. Migrate	A working hybrid X25519 + ML-KEM-768 handshake and ML-DSA-65 signatures beside RSA and ECDSA.

Screenshots

Our finding

We measured the effective error per two-qubit gate on three IBM quantum computers from every Grover run. The best qubits on all three agree at about 3 × 10⁻³, three times the error rate the published RSA-2048 estimate assumes. Carried through a surface-code model, breaking RSA-2048 on today's qubits would take roughly 4 million physical qubits rather than under 1 million. On one machine the compiler's default qubits measured 1.6 × 10⁻²: above the error-correction threshold, where no amount of hardware helps. See Methods and Hardware results.

Where to go next

• User guide: install, run the dashboard, every CLI option, configuration and deployment.
• Methods: how break years, risk levels, the hardware tests and the error-rate measurement work.
• Hardware results: every IBM job ID, with 95% intervals and baselines.
• Architecture: how the code is organised and why.
• API reference: use qday from Python.

Built by Team Straw Hat Pirates (Raunak Kumar Gupta, Krish Jain) for the VTU-QFF Hackathon, Qiskit Fall Fest 2026.