v0.31 ships the KuzuDB-backed knowledge graph, EU AI Act / GDPR governance exports, and the nova-worm-conformance test suite for object storage compliance.
v0.31 is the largest release since the initial capsule format stabilised. Three major features ship together: the knowledge graph backend, governance compliance exports, and a standalone WORM conformance test suite.
All capsule relationships — asset versions, run lineage, model call chains — are now queryable via a KuzuDB-backed graph. The dashboard includes graph visualisation and Cypher query panels. A 10M-edge benchmark confirms the backend handles realistic production-scale capsule histories.
Risk classification and audit profiles are now available for EU AI Act, NIST AI RMF, GDPR, SOC 2, and NIS2. These are evidence exports for compliance workflows — not legal guarantees. Regulatory fit requires independent review by someone qualified to give it. Cost reporting and per-run audit trails ship alongside them.
A standalone Python package that verifies object storage backends (S3, MinIO, Ceph, Azure Blob) meet WORM requirements for SEC 17a-4, MiFID II, CFTC 1.31, and FINRA 4370. Ten mandatory test cases covering immutability, retention locks, legal holds, and lifecycle exemptions.
$ pip install "git+https://github.com/MSKazemi/novafabric#subdirectory=packages/nova_worm_conformance"
$ nova-worm-conformance run --backend s3 --bucket my-capsule-store
✓ 10/10 conformance tests passed
report: worm-report-2026-05-20.jsonCorrection, 2026-08-08: this post originally gave the install command as
pip install nova-worm-conformance. That package was never published to PyPI, so the
command could not work for anyone who tried it. The code is real and lives in the main
repository under packages/nova_worm_conformance; the command above installs it from
there. Apologies to anyone who hit the error — if the package is published standalone
later, the short form will come back.