Skip to content
Get packs

Guide · Agent skill

How to use the Cirq quantum circuits agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The Cirq skill gives an AI agent working knowledge of Google's open-source Cirq library, so it can write quantum circuit code, parameter sweeps, noise models and hardware-targeting scripts that follow Cirq's conventions. This guide explains what the skill does, how to install it, how to ask for things in plain language, and what to check before running anything. You do not need prior experience with agent skills to follow it.

Cirq Quantum CircuitsK-Dense-AI/scientific-agent-skills · MIT

Quick start

  1. 1Download and unzip. Download the skill ZIP from this page and unzip it so you have a cirq folder.
  2. 2Place the folder. Move the cirq folder into ~/.claude/skills/ for personal use or into .claude/skills/ in your project. Alternatively, upload the ZIP under Settings → Capabilities → Skills in the Claude apps.
  3. 3Install Cirq. Run uv pip install "cirq-core==1.7.0" on Python 3.11 or newer, and add any hardware package you need on the same version.
  4. 4Ask in plain language. Describe the circuit or experiment you want and mention Cirq, so the agent loads the skill.
  5. 5Simulate locally. Have the agent run the circuit on a local simulator and check the histogram or expected values.
  6. 6Review before hardware. Read any script that submits jobs, confirm the target and shot budget, and only then allow the cloud run.

What the Cirq skill makes the agent do

Cirq is Google Quantum AI's Python framework for designing quantum circuits, simulating them on an ordinary computer, and submitting them to quantum hardware. The skill is a set of instructions and reference guides that the agent reads when your request matches its description. It covers circuit building, simulation, circuit transformation, hardware integration, noise modeling and experiment design.

The skill itself is documentation and examples. The agent still runs the Python code on your machine when you ask it to.

  • Build circuits with qubits, gates and measurements
  • Run simulations and parameter sweeps
  • Model noise and design characterization experiments
  • Compile and optimize circuits for hardware gatesets
  • Prepare scripts for Google, IonQ, AQT, Pasqal and Azure Quantum targets

Install the Cirq skill from a ZIP file

Download the skill ZIP from this page and unzip it. Then place the resulting cirq folder in one of the locations below. Use the personal location to make the skill available across your projects, or the project location to limit it to one codebase.

You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

  • Personal: unzip into ~/.claude/skills/
  • Project: unzip into .claude/skills/ inside your project folder

Install the Cirq packages the examples need

The skill's examples target Cirq 1.7.0 on Python 3.11 or newer. Install the core package, and add the hardware package only for the provider you plan to use. Keep all cirq-* packages in one environment on the same release.

Azure Quantum needs a separate environment, because its supported stack uses Cirq 1.6.1. Do not force Cirq 1.7.0 into that environment.

  • cirq-core is required for local circuits and simulation
  • ply is optional, for OpenQASM import
  • cirq-google is for Google Quantum Engine and needs approved Google Cloud project access
  • cirq-ionq, cirq-aqt and cirq-pasqal are for their respective providers
uv pip install "cirq-core==1.7.0" # Optional OpenQASM import dependency uv pip install ply

Trigger the skill with natural requests

You do not need a special command. Describe the quantum task in plain language and mention Cirq or the kind of work involved. The skill's description points to Google hardware, noise-aware circuit design and characterization experiments, so requests that match those areas are most likely to load it.

If your task belongs to another toolchain, the skill says so. For IBM hardware it recommends qiskit, for quantum machine learning with autodiff it recommends pennylane, and for physics simulations it recommends qutip.

  • Write a Cirq circuit that prepares a Bell state and measures both qubits
  • Add depolarizing noise to this circuit and compare the histograms
  • Sweep a rotation angle from 0 to pi and plot the measured probabilities
  • Help me target a Google Quantum AI processor with this circuit, checking the gateset first
  • Write a Cirq script that compiles this circuit for the IonQ target

Check the code before any cloud or hardware run

Local simulations are cheap and safe to repeat. Cloud and hardware runs are not. Cloud execution uses your provider credentials and network access, may incur costs, and only works on targets assigned to you. The skill's hardware guide states that its cloud snippets are illustrative and were not run as authenticated end-to-end tests.

Ask the agent to show you the script, explain which target and how many repetitions it uses, and confirm that it simulates the circuit locally first. Only then run anything that submits jobs.

  • Confirm the provider, target device and shot count before submission
  • Simulate small circuits locally to check results first
  • Verify that the qubits you choose are physically connected on the device
  • Keep credentials out of the code you share

What people use it for

  • Teaching a first Bell-state circuit with measurement histograms
  • Running a parameter sweep over a rotation angle and plotting the results
  • Adding noise models to see how a circuit degrades before hardware runs
  • Compiling a circuit to an IonQ, AQT or Pasqal gateset for a provider script
  • Building a variational algorithm template with a classical optimizer

Frequently asked questions

Is the Cirq skill free to use?

The skill is released under the MIT licence, as listed on this page. Cirq itself is Google's open-source Python library. Cloud hardware access is separate: it depends on your provider accounts and may incur charges.

What do I need to run the Cirq skill?

You need Python 3.11 or newer and the cirq-core 1.7.0 package for local work. Hardware use requires the matching provider package and access. For Google hardware that means approved Google Cloud or Quantum Engine project access.

How do I install the Cirq skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or into .claude/skills/ in a project. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.

Can I use Cirq for IBM quantum computers?

The skill recommends qiskit for IBM hardware, so you should use that toolchain instead. Cirq is not the right choice for IBM device work.

Why does the skill need network access?

Network access is only needed for cloud execution, such as submitting jobs to Google Quantum Engine or partner providers. Cloud runs need your provider credentials and assigned targets. Local simulation does not require it.

Why won't my Cirq code run on Azure Quantum with Cirq 1.7.0?

The Azure path uses its own environment with qdk and azure-quantum and a supported Cirq 1.6.1 stack. Install Azure packages in a separate environment rather than forcing Cirq 1.7.0 into it.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use the Cirq skill if you are learning or building quantum circuits in Python and want an agent that knows Cirq's conventions, noise modelling and hardware targets. Pick qiskit for IBM hardware, pennylane for quantum machine learning with autodiff, or qutip for production physics simulations. Review any script that submits cloud jobs before you run it.

Download the skill