Evidence

Verbatim from the full PDF text: “Only eight of 26 respondents (31%) reported using quantum specific testing tools.” Stated as Finding 2, Section 5.2, p.18. Survey of 26 analyzed respondents (from 1,397 industry/government + 75 academic invitees), plus 4 follow up interviews, fielded via Qualtrics 2024-05-20 to 2024-08-05, IRB approved.

Zappin, Stalnaker, Chaparro, Poshyvanyk, “Challenges and Practices in Quantum Software Testing and Debugging: Insights from Practitioners,” arXiv:2506.17306

Authors themselves flag N=26 as a generalizability limitation (Section 9.2). Paper explicitly calls for, but does not build, “CI/CD pipelines tailored to hybrid systems”, evidence of need, not of prior art absence.

Qiskit’s official ecosystem CI template ships three real, verified workflow files (test_latest / development / minimum_versions.yml), reused across ~8 Qiskit ecosystem repos. Qiskit core’s own qpy.yml performs genuine cross Qiskit version regression testing for QPY serialization compatibility.

Qiskit ecosystem CI template; Qiskit QPY compatibility harness

These templates assert pass/fail of existing tests at each version tier, they do not diff behavior between versions themselves.

No specific arXiv IDs were supplied to check against, so the original “two papers” could not be confirmed or denied directly. Independent search reconstructed the most material candidates: QUTest (arXiv:2605.19736) already implements cross Qiskit version regression testing with CI compatible (JUnit/xUnit) output for GitHub Actions, but does not integrate with pytest, and supports only Qiskit today.

Campos, “QUTest: A Native Testing Framework for Quantum Programs,” arXiv:2605.19736

The single most material finding of the whole research phase: it directly narrows the novelty claim. What remains open per QUTest’s own stated limitations: pytest native integration, cross SDK testing, automated (not hand written) regression detection.

Three real, installable pytest plugins exist (pytest-quantum, qtest-quantum, qc-assert) but do framework to framework (not version to version) equivalence or single run assertions only. All are under 9 months old, single maintainer, 0-2 GitHub stars.

See the Prior Art matrix for the full inspection table

None execute a test suite against multiple installed versions of the same SDK and auto diff results.

quantum-transpiler-regression-testing (“cart”), independently verified via GitHub repo and Zenodo DOI resolution, found that ~38% of real Qiskit transpiler bug fixes are regressions invisible to black box equivalence oracles. Scope: single SDK, CLI based, brand new (0 stars). No cross SDK regression/equivalence tool was found anywhere, the closest cross SDK system, Benchpress (C14), is a performance benchmark with no detection capability.

furqan-nr/quantum-transpiler-regression-testing; Zenodo 10.5281/zenodo.21020113

Treat as directly relevant, possibly competing/complementary prior art.

Benchpress is a pytest based benchmarking suite of 1,000+ tests executed across 8 SDKs (Qiskit, Braket, Cirq, Tket, BQSKit, Staq, pyqpanda3, and the Qiskit IBM transpiler) on circuits up to 930 qubits and ~1M two qubit gates. It proves the pytest native, cross SDK orchestration pattern works at scale, the two pieces of the proposed gap it covers. It does not cover the other two: every test yields capability/performance results (pass/skip/fail/xfail plus timings via pytest-benchmark); there is no circuit equivalence checking, no cross version regression detection, and no reusable GitHub Actions packaging.

Qiskit/benchpress (README, gym layout); “Benchmarking the performance of quantum computing software,” arXiv:2409.08844, publ. Nat. Comput. Sci. 5, 427-435 (2025)

Apache-2.0, ~155 stars, 323 commits. Found 2026-08-21, after the original 2026-08-16/20 sweep, added under the same verify by opening rule. Refines C10/C13: the narrowed gap statement stands, but “no cross SDK tool of any kind” is now “no cross SDK regression/equivalence tool.”

MQT QCEC is a mature equivalence checking library (decision diagram, alternating DD, simulation based falsification, ZX calculus engines) for Qiskit circuits and OpenQASM files. Its own CI matrix is OS/architecture/compiler only, no SDK version matrix, no cross SDK version regression harness, no CI/CD product for arbitrary user projects.

munich-quantum-toolkit/qcec (README, CI workflow, verification handbook)

117 GitHub stars, 1,516+ commits, MIT license. Citation: Burgholzer & Wille, “QCEC: A JKQ tool for quantum circuit equivalence checking,” Software Impacts, 2021.

MQT Debugger is an assertion based, simulation driven debugger for locating errors within a single quantum program run, with a Debugger Adapter Protocol (DAP) server for IDE integration. Its GitHub Actions workflows are standard CI for its own codebase, not a feature for testing other projects across SDK versions.

Rovara, Burgholzer, Wille, “A Framework for Debugging Quantum Programs,” arXiv:2412.12269

21 stars, 4 forks, 496 commits, MIT license. arXiv:2412.12269 submitted 2024-12-16.

“Q-Trace” as a named quantum debugging tool does not exist. Other real systems verified. QMon, TraceQ, the Microsoft QDK Trace Simulator, CUDA-Q statistical assertions, Proq, Bloq/AutoBloq, MorphQ, all operate at the single circuit, single run fault localization or monitoring level. None perform cross version or cross SDK regression comparison; none integrate with pytest or GitHub Actions.

See the Prior Art matrix for the full per system table

MorphQ (Qiskit testing Qiskit) flagged for cross check but not deeply verified in this pass.

Bugs4Q (36 bugs in the 2021 preprint; 42 in the DOI resolved JSS 2023 version) is described by independent replication work as “a widely used dataset.” Critically, arXiv:2606.27124 ran 37 Bugs4Q artifacts across 21 Qiskit core library versions (77,700 executions), found reproducibility collapsed from 62.2% on v0.20.1 to 16.2% on v2.3.1, and released a patched fork “Bugs4Q-Robust” that restores it to 78.4%.

Bugs4Q (arXiv:2108.09744, JSS vol. 205 2023); cross version replication (arXiv:2606.27124)

No single corpus is a universal standard (unlike Defects4J for Java), the narrowest defensible claim is that existing corpora are not packaged as ready to use pytest/CI fixtures. The replication authors note most failures needed source code migration (import paths, API calls), not just dependency pinning, version drift in quantum SDKs is a code level, not packaging level, problem.

GitHub repository search for “quantum equivalence checking” returned only unrelated SAT solver student projects whose legitimacy was not independently verified. No other academic or industry open source equivalence checker with CI integration was found beyond MQT QCEC, QUTest, and the “cart” pilot.

gh search repos “quantum equivalence checking”

Absence of evidence after search, not proof of nonexistence.

Generic autonomous AI testing agents exist abundantly for classical software QA with no quantum angle. The closest quantum specific agentic system, QBugLM, is genuinely agentic and quantum specific but has no CI/CD integration, no cross SDK version regression detection, and no demonstrated state awareness.

QBugLM (arXiv:2606.07314); general web search for autonomous quantum testing agents

Explicitly marked NOT FOUND rather than FALSE, absence of evidence after a reasonably broad search, not proof of nonexistence. Least load bearing part of the original idea.