Changelog

All notable changes will be documented here. The project follows Keep a Changelog structure and uses semantic versioning for public releases.

Unreleased

accept documented dotted CLI rule IDs while still rejecting unknown rule fields;
retain input and parser failures in CLI/direct CliAnalyzer results so corrupt archives or malformed classes cannot silently produce passing partial checks; and
reject incomplete graph imports by default before writing graph output.

align the project license with the ArchUnitEverything family by adopting the MIT License.
update repository ownership, maintainer metadata, and documentation links after the GitHub transfer to Lukas Niessen while retaining the published Maven Central coordinates.

identify the published line consistently as a public beta;
add a feature-oriented user guide, complete CLI configuration reference, support policy, issue forms, pull-request checklist, and code of conduct; and
introduce the Java orange-and-blue AU family logo plus a dedicated GitHub social preview.

deterministic class-file import and immutable Java architecture models;
selectors, graph projections, architecture rules, presets, metrics, and reviewed baseline values;
CLI, JUnit Platform, Maven/Gradle bridge, and multi-format reporting integrations;
bounded importer/cache/rendering security controls and an adversarial corpus;
pinned open-source performance baselines and semantic regression snapshots;
reproducible release-candidate packaging, source/Javadoc artifacts, and a Maven consumer example;
strict bounded baseline JSON ingestion, spreadsheet-safe CSV, safe/trusted regex separation, and complete package-info.class graph identities;
Maven Central-ready coordinates, metadata, signing policy, bundle generation, and tag-gated user-managed staging workflow.

defensive handling for file-system roots whose Path.getFileName() is absent;
explicit cache-output stream ownership and pre-construction validation for execution errors; and
defensive copies at the class-file parser backend seam.

JDK 25 is required to run the library.
The public API is provisional before 1.0 and is not ArchUnit-compatible.
The performance corpus is regression evidence, not an absolute scalability claim.
Unrestricted regex and lossless non-neutralized CSV require explicitly named trusted APIs.