Skip to content
Back to skills

Io Data Logging

ASecurity

"Use PennyLane circuit import/export, dataset loading, debugging

  • 247 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added September 8, 2026
testingpythonnodedebuggingapi

Works with

  • api

Security analysis

A100/100

Pro scans all 5 files and shows the line behind each finding

Scanned September 8, 2026

npx -y skills add VectorSpaceLab/AREX-Skill --skill io-data-logging --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Io Data Logging?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Io Data Logging
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vectorspacelab-io-data-logging/badge)](https://www.skillsdirectory.com/skills/vectorspacelab-io-data-logging)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: io-data-logging
description: "Use PennyLane circuit import/export, dataset loading, debugging
  snapshots, logging, pytrees, and concurrency support workflows."
metadata:
  disco-role: operating
disable-model-invocation: true
license: Apache 2.0
---

# I/O, data, debugging, and logging

Use this sub-skill when a task involves importing/exporting circuits, loading PennyLane datasets, debugging circuit execution, configuring logging, or using support utilities such as pytrees and concurrency.

## Read first

- [`references/api-reference.md`](references/api-reference.md): verified signatures and module responsibilities for I/O, data, debugging, and logging.
- [`references/workflows.md`](references/workflows.md): OpenQASM export, converter dependency checks, dataset loading, snapshots, and logging recipes.
- [`references/troubleshooting.md`](references/troubleshooting.md): optional converter imports, remote dataset/cache failures, debug measurement issues, and logging config mistakes.
- [`scripts/io_data_smoke.py`](scripts/io_data_smoke.py): safe CPU smoke for OpenQASM export and importable data/debugging/logging modules.

## Route within this sub-skill

- **Circuit export:** use `qp.to_openqasm(circuit, wires=None, rotations=True, measure_all=True, precision=None)` for OpenQASM 2 output where supported.
- **Circuit import/conversion:** use `qp.from_qasm`, `qp.from_qasm3`, `qp.from_qiskit`, `qp.from_qiskit_noise`, `qp.from_qiskit_op`, `qp.from_pyquil`, `qp.from_quil`, `qp.from_quil_file`, `qp.FromBloq`, and `qp.to_bloq` only after checking optional dependencies.
- **Datasets:** use `qp.data.load(...)` and dataset namespace utilities; treat remote downloads/cache mutation as opt-in behavior.
- **Debugging:** use `qp.snapshots`, `qp.breakpoint`, `qp.debug_expval`, `qp.debug_state`, `qp.debug_probs`, and `qp.debug_tape` for circuit introspection.
- **Logging:** use `pennylane.logging` and its TOML configuration surfaces when instrumenting PennyLane behavior.
- **Support utilities:** use `pennylane.pytrees` and `pennylane.concurrency` for advanced integration tasks after confirming exact APIs.

## Boundaries

- Circuit execution and measurements belong to [`../circuits-devices/SKILL.md`](../circuits-devices/SKILL.md).
- Operator transformations and decomposition belong to [`../operators-transforms/SKILL.md`](../operators-transforms/SKILL.md).
- Dataset-backed scientific qchem/application workflows belong to [`../applications-qchem-resource/SKILL.md`](../applications-qchem-resource/SKILL.md) when the domain module owns semantics.
- Source-checkout test/lint/development policy belongs to [`../repo-development/SKILL.md`](../repo-development/SKILL.md).

## Minimal OpenQASM export

```python
import pennylane as qp

dev = qp.device("default.qubit", wires=2)

@qp.qnode(dev)
def circuit(theta):
    qp.RX(theta, 0)
    qp.CNOT([0, 1])
    return qp.expval(qp.Z(1))

print(qp.to_openqasm(circuit)(0.2))
```

If a converter asks for Qiskit, PyQuil, OpenQASM3, or Qualtran, install and verify only that converter dependency.

## Verification cues

State whether the workflow is local-only or requires network/cache/external converter packages. For debug/logging answers, include a tiny QNode and expected printed/debug signal rather than only prose.

Files in this skill

  • SKILL.md3.2 KB
  • references/api-reference.md2 KB
  • references/troubleshooting.md2.1 KB
  • references/workflows.md2.2 KB
  • scripts/io_data_smoke.py675 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.