# qiskit

> IBM quantum computing framework — use when targeting IBM Quantum hardware, working with Qiskit Runtime for production workloads, or needing IBM optimization tools for quantum error mitigation and enterprise quantum computing.

**Use case**: Build and run quantum circuits on IBM Quantum hardware

**Canonical URL**: https://agentcookbooks.com/skills/qiskit/

**Topics**: claude-code, skills, science, quantum-computing

**Trigger phrases**: "quantum circuit with Qiskit", "IBM Quantum", "quantum gate", "run on quantum hardware", "Qiskit Runtime"

**Source**: [K-Dense AI](https://github.com/K-Dense-AI/scientific-agent-skills/tree/main/scientific-skills/qiskit)

**License**: Apache-2.0

---

## What it does

`qiskit` is a Claude Code skill from K-Dense AI's [scientific-agent-skills repo](https://github.com/K-Dense-AI/scientific-agent-skills). It turns Claude into a Qiskit expert covering quantum circuit construction (`QuantumCircuit`), transpilation for specific IBM backend topologies, simulation (`AerSimulator`, `StatevectorSimulator`), Qiskit Runtime for hardware execution, quantum error mitigation (ZNE, PEC), variational algorithms (VQE, QAOA), and the Qiskit ecosystem (qiskit-nature, qiskit-optimization, qiskit-machine-learning).

A session produces Qiskit Python code: a quantum circuit built from the required gates, transpiled to the target backend's native gate set, and either simulated or queued for hardware execution via Qiskit Runtime.

## When to use it

Reach for it when:

- You're targeting IBM Quantum hardware specifically and need transpilation, error mitigation, and runtime job management
- You're implementing variational quantum algorithms (VQE for chemistry, QAOA for combinatorial optimization) on IBM's stack
- You're using Qiskit's ecosystem packages (qiskit-nature for quantum chemistry, qiskit-optimization for combinatorial problems)

When *not* to reach for it:

- Google hardware (Cirq is the native framework)
- Gradient-based quantum ML across multiple hardware backends — use `pennylane`
- Open quantum systems simulations — use `qutip`

## Install

Copy the `SKILL.md` from K-Dense AI's [qiskit folder](https://github.com/K-Dense-AI/scientific-agent-skills/tree/main/scientific-skills/qiskit) into `.claude/skills/qiskit/` in your project. Install via `pip install qiskit qiskit-aer qiskit-ibm-runtime`. An IBM Quantum account is required for hardware access.

Trigger phrases: "quantum circuit with Qiskit", "IBM Quantum", "quantum gate", "run on quantum hardware".

## What a session looks like

A typical session has three phases:

1. **Circuit design.** Describe the quantum algorithm or the circuit structure. Claude builds the `QuantumCircuit` with the appropriate qubit count, gates, and measurement operations, then draws the circuit for visual verification.
2. **Backend selection and transpilation.** Claude transpiles the circuit to the target backend's native gate set and qubit coupling map — essential for hardware execution but optional for simulation. Simulator selection (AerSimulator for noise modeling, StatevectorSimulator for exact state) is set based on the task.
3. **Execution and analysis.** For simulation, results are retrieved immediately. For hardware via Qiskit Runtime, the job is submitted and the skill includes code to monitor and retrieve results asynchronously. Measurement outcomes are parsed into probability distributions or expectation values.

## Receipts

**Where it works well:**
- Bell state and GHZ circuit construction for testing quantum hardware connectivity — standard circuits that Claude generates correctly with measurement calibration
- VQE for small molecule ground state energies (H2, LiH) — the qiskit-nature integration handles the Hamiltonian mapping and ansatz selection cleanly for textbook molecules

**Where it backfires:**
- Noisy intermediate-scale quantum (NISQ) hardware results vary across execution runs — Claude's error mitigation code helps but doesn't eliminate the fundamental noise floor
- Hardware queue times on IBM Quantum free tier can be hours; Claude doesn't proactively suggest simulator-first development workflows for users unfamiliar with the queue dynamics

**Pattern that works:** always simulate locally with AerSimulator (optionally with a noise model from the target device) before submitting to real hardware — catching circuit bugs before queuing saves significant wait time.

## Source and attribution

Originally authored by [K-Dense Inc.](https://github.com/K-Dense-AI). The canonical SKILL.md lives in the [`qiskit` folder](https://github.com/K-Dense-AI/scientific-agent-skills/tree/main/scientific-skills/qiskit) of their public scientific-agent-skills repository.

License: Apache-2.0. Install, adapt, and redistribute with attribution preserved.

This page documents the skill from a practitioner's perspective. For the formal spec and any updates, defer to the source repo.