Nuclei developer docs

Nuclei is the open-source quantum workspace — learn it, research it, run it on real hardware, all local and private. For developers it is four things: a scriptable simulation kernel behind a plain WebSocket JSON protocol, a multi-framework abstraction that normalizes Qiskit, Cirq, CUDA-Q, and Q# into one CircuitSnapshot format, a gateway that submits the same circuits to real quantum hardware, and an AI layer (Dirac) that calls api.anthropic.com directly with a user-supplied key — no Nuclei server in the loop. If you want to use Nuclei, download the app or try it in the browser — this site is for people building on Nuclei.

What can I run?

	Supported today
Languages	Python · Q#
Frameworks / SDKs	Qiskit · Cirq · CUDA-Q · Microsoft QDK
Simulators	Aer · `cirq.Simulator` · CUDA-Q CPU/GPU · QDK sparse simulator
Hardware	IBM · IonQ · Quantinuum · AWS Braket · Azure Quantum · NVIDIA CUDA-Q GPU

Google Quantum AI is scaffolded only: a provider stub exists in the kernel but always refuses to connect, and its Launch-portal tile is disabled (“Research access only”). Don’t build against it yet.

Find your path

Inspect — see exactly what the compiler does to your circuit, pass by pass, with the Transpiler Explorer.
Integrate — drive the kernel from your own client, grader, or notebook: Kernel API overview.
Extend — add a framework adapter, hardware provider, or contribute to the IDE: framework adapters, hardware providers, contributing.
Evaluate — understand how the pieces fit before committing: architecture and desktop vs web.