Skip to content

Security: hinanohart/quantum-flex

SECURITY.md

Security policy

Reporting a vulnerability

quantum-flex is a research-and-visualization library. Most of the surface area touches local files, matplotlib, and outbound IBM Quantum API calls. Even so, several classes of issue do warrant a coordinated disclosure:

  • Code paths that could exfiltrate an IBM Quantum API token from a user's environment (e.g. a logger, a JSON writer, or an exception path that leaks QISKIT_IBM_TOKEN to disk or stdout).
  • Deserialization issues in RunRecord.load or in the data/runs/ parsers.
  • Supply-chain issues in our pinned dependencies.
  • Anything that lets a malicious notebook escape the user's local workspace.

Please report these privately to hinanohart@proton.me. We will:

  1. Acknowledge within 72 hours.
  2. Confirm the issue and prepare a fix in a private branch.
  3. Coordinate a disclosure window (default 90 days, shorter if active exploitation is observed).
  4. Publish a GitHub Security Advisory and credit the reporter (unless they prefer to remain anonymous).

For non-security bugs, please open a regular GitHub issue.

What we do NOT consider a vulnerability

  • Numerical noise in IBM Quantum results — that is the physics, not a bug.
  • Failed experiments under experiments/_wip/. Those are deliberately preserved failures (see the failure-museum policy in CONTRIBUTING.md).
  • Differences between Aer simulation and real-hardware execution.
  • Notebook cells that render slowly on a low-memory machine.

Scope

The scope of this policy is the contents of this repository: source code under src/, tests, build configuration, GitHub Actions workflows, JSON schema parsers, and committed datasets.

We do not provide security guarantees for third-party Qiskit plugins or IBM Quantum services themselves — those have their own disclosure channels:

There aren't any published security advisories