Skip to content

Quantum Kuramoto audit usage

Generate an import audit, boundary review and API contract from explicitly selected inputs. Each review and contract records the actual input path and SHA-256. The date in the output filename identifies the original report format; it does not establish when the selected source was qualified.

Run from the repository root with the project dependencies installed. Select an output directory outside the archived data/s6_quantum_kuramoto_split reports:

python scripts/audit_quantum_kuramoto_split.py \
  --source-root src/scpn_quantum_control \
  --out-dir /tmp/quantum-kuramoto-audit \
  --doc-path /tmp/quantum-kuramoto-audit/audit.md
python scripts/export_quantum_kuramoto_boundary_review.py \
  --audit-path /tmp/quantum-kuramoto-audit/quantum_kuramoto_split_audit_2026-05-07.json \
  --out-dir /tmp/quantum-kuramoto-audit \
  --doc-path /tmp/quantum-kuramoto-audit/review.md
python scripts/export_quantum_kuramoto_api_contract.py \
  --review-path /tmp/quantum-kuramoto-audit/quantum_kuramoto_boundary_review_2026-05-07.json \
  --out-dir /tmp/quantum-kuramoto-audit \
  --doc-path /tmp/quantum-kuramoto-audit/contract.md

The audit reads the selected physical Python sources, including the root Rust accelerator entry point. Relative imports are resolved against their containing module. Existing accelerator compatibility aliases are checked against their importable canonical provider and record that provider's physical source digest. This requires the canonical provider to be available in the Python environment. AST classifications describe source dependencies; they do not establish runtime correctness, scientific fidelity or package publication readiness.

For Python consumers, audit_module_source(module, path) audits one physical module and build_split_audit(source_root=...) collects the source inventory. write_split_audit(payload, json_path=..., doc_path=...) writes JSON and Markdown and returns their byte digests. build_boundary_review(audit, source_audit=path) and build_api_contract(review, source_review=path) bind the supplied objects to the declared files. A file whose JSON content differs from the supplied object raises ValueError. An in-memory object supplied without a source path remains explicitly unbound. Omitting the object uses the archived default input and records its provenance.

The reports retain conservative package and publication holds. Generating them does not create or authorize a new package. The dated archived reports remain historical inputs unless explicitly selected as output destinations.