v1.3.1 — Consistent document layouts and mobile reading

Reference pages now share a consistent document layout. Headings align with the reading column; desktop pages have a sticky contents list, and mobile pages use a compact native disclosure control.

The specification replaces the wide question table with numbered steps, stacks mapping fields on phones, and adds labelled examples with copy buttons. Project status is shorter and easier to scan. The gallery spacing and English code direction on RTL pages are also corrected.

Classification criteria, decision branches and pending human-review status are unchanged. The full build, data/schema checks, MCP protocol tests and Chrome checks passed. The browser checks cover 22 languages, 32 wizard paths, responsive layouts, keyboard use, themes, RTL, no JavaScript and automated WCAG A/AA checks.

Attached: vocabulary, both JSON Schemas and 54 SVG marks in ai-usage-scale-v1.3.1.zip (CC0); the local MCP server in usagescale-mcp-1.3.1.tgz (MIT). The MCP package remains distributed through source and release archives.

v1.3.0 — Typographic reference and verified tooling

The website is now a typographic reference for the six definitions, with a practical declaration flow and a public project-status page. Core pages remain available in 22 languages.

This release removes the maintainer-only signature list, unsupported adoption examples, assumed Level 3 website metadata and automatic human-approval claims. README, policy references, privacy statements and licence notices now reflect the project's actual scope and evidence.

• JSON-LD uses an explicit project extension. Both vocabulary and declaration schemas are available and included in the download.
• Astro and MCP dependencies are updated. The release audit reported zero known npm vulnerabilities in both packages.
• MCP validates inputs, includes the missing classifier in its package and exposes the provisional Level 3 review boundary.
• CI now checks all 352 generated pages, the six reachable levels, all 32 wizard paths, responsive layouts, keyboard behavior, no-JavaScript behavior and automated accessibility checks. CodeQL and Dependabot are enabled.

Compatibility and review

Classification criteria and decision branches are preserved from v1.2.0. The editorial sentence suggesting that disclosure makes a Level 5 work legitimate is removed in all languages. Signing URLs redirect to project information.

The W3C CG mapping candidates follow its 9 October 2026 three-value text report. w3cAiDisclosure is now null for Levels 1 and 2: text scope and actual review must be assessed separately. Generic numeric-level snippets omit a CG declaration. Suggested mappings do not imply institutional acceptance or Google Search support.

The Level 3 definition requires meaningful human review, while its existing decision shortcut does not ask about review. Its automated result remains provisional. Human editorial review of the revised website and fluent review of the 21 translations remain pending.

Downloads

• ai-usage-scale-v1.3.0.zip: the CC0 vocabulary, both JSON Schemas and all 54 SVG marks, with the licence-scope document.
• usagescale-mcp-1.3.0.tgz: the MIT local MCP server and bundled vocabulary. It is not published on npm. After downloading, install with npm install -g ./usagescale-mcp-1.3.0.tgz; the command is usagescale-mcp.

See the changelog, October audit and public standards/pilot record.

v1.2.0

The marks (CC0) and levels.json (CC0) for this version are attached.

A major version bump means a level's meaning changed — which invalidates
declarations already published against it. See CHANGELOG.md.

What's Changed

• Release v1.2.0: EU AI Act mapping, 242 translation fixes, wider document by @karauda in #10

Full Changelog: v1.1.1...v1.2.0

v1.1.1

The marks (CC0) and levels.json (CC0) for this version are attached.

A major version bump means a level's meaning changed — which invalidates
declarations already published against it. See CHANGELOG.md.

Full Changelog: v1.1.0...v1.1.1

v1.1.0

The marks (CC0) and levels.json (CC0) for this version are attached.

A major version bump means a level's meaning changed — which invalidates
declarations already published against it. See CHANGELOG.md.

Full Changelog: v1.0.0-draft...v1.1.0

v1.0.0-draft

The marks (CC0) and levels.json (CC0) for this version are attached.

A major version bump means a level's meaning changed — which invalidates
declarations already published against it. See CHANGELOG.md.

Full Changelog: https://github.com/karauda/ai-usage-scale/commits/v1.0.0-draft