日本語

• Separate the shared Markdown foundation into Syntamark, using the published Syntamark 1.0.0 runtime and compiler-source artifacts from one verified pair.
• Replace the Testing Document getter with LithoSharp-owned read-only HTML snapshots and queries. This is a source and binary breaking change.
• Add optional analyzers for definite misuse of registered APIs, supported HTML interpolation/URLs, and references to verified catalogs. Unknown and unsupported cases do not become complete proofs.
• Add static-input and trusted-catalog preflight, plus the read-only migrate testing --dry-run report, to the Tool. See the 2.0 migration notes for execution boundaries and examples.

• Editor inspection shares one analysis with the CLI: opt-in Markdown compatibility advisories (LIT003/LIT004/LIT005), project-aware snapshots with route candidates, and an analysis-only MDX worker session that never loads plugins, bundles, or runs SSR.
• Minimal stdio language server plus a VS Code 0.1.0 shell: project detection without MSBuild evaluation, limited workspace trust, build/serve control with structured shutdown, debounced diagnostics, symbols, and a real-output preview.
• Docusaurus migration gains a classified route oracle, an explicit normalized page-set comparison, and component functional-change classifications; a pinned Prettier/Jest/Docusaurus corpus reproduces conversion end to end.
• No incompatible public signatures. Markdown compatibility stays limited to the documented subset. See the 1.1 features and upgrade notes.

• Replace the runtime Markdig dependency with the Litho Markdown compiler. HTML, headings, links, assets and search text share a parsed document; persistent parse records are verified before reuse.
• Add stable document inspection, workspace snapshots, front matter schemas, navigation lookups, structured diagnostics and versioned CLI/serve output.
• Add Docusaurus migration analysis, conversion into a separate directory, route comparison and fingerprinted suggested edits. Configuration is never executed during migration; unsupported constructs require the documented steps.
• Extend the supported MDX component subset and preserve original diagnostic lines when processing mdx-code-block fences.
• Protect inspection, schema and navigation snapshots from mutation, decode heading text once in tables of contents, and reject symbolic migration destinations.

Public API additions preserve the existing signatures. Markdown compatibility is limited to the documented subset: footnotes, definition lists and other listed Markdig extensions are not implemented. Regenerate sites and review content warnings before upgrading. See the Docusaurus migration guide and tooling compatibility policy.

• Refresh the built-in Docs and Blog layouts, typography, spacing and responsive navigation.
• Add light/dark palettes with system preference and saved reader selection. SiteThemeOptions.EnableThemeSwitching defaults to true; set it to false for fixed light colors without the toggle or preference scripts. Fixed dark colors are available through AdditionalCss.
• Improve skip links, keyboard focus, Escape handling and navigation without JavaScript.
• Update the shared NuGet icon and add sample screenshots to the READMEs.

Existing public signatures and routes are preserved. Generated HTML, CSS and JavaScript change; review custom CSS after regeneration. See the theme configuration.

• Typed Markdown, YAML, JSON and CSV collections, routes, references, schemas and source generation with build-time diagnostics.
• Dependency graphs, persistent caches, invalidation reports, artifact ownership, staged output transactions, quality checks, redirects and fingerprinted assets.
• Optional image transformations, .NET testing APIs, CLI/site factories and four project templates: Docs, Blog, empty C# layout and MDX.
• Optional MDX 3 and React with build-time HTML, page hydration, explicit islands and selective activation strategies.
• Versioned and multilingual Docs, multiple collections, Blog/Pages, local search, supported Docusaurus aliases, XML/OpenAPI reference pages and exact xrefs.
• Progressive browser navigation, opt-in offline support, sandboxed live code and consent-controlled analytics.

• Publication conditions now filter legacy posts before rendering and derived artifact generation. Output routes and ownership receive stricter validation.
• Missing social images no longer produce dangling image metadata. Text output normalization and content-based search fingerprints improve reproducibility.
• Core adds AngleSharp 1.7.3. Six optional package IDs join LithoSharp; existing Markdown applications do not need to adopt them.

• The measured static-page fixture reduced JS from 449,585 to 15,817 bytes and hydration roots from one to zero with selective hydration.
• Markdown 100/1,000-page repeated medians stayed within 10% of the benchmark baseline for time, allocation and resident-memory increases.
• Large MDX incremental rendering can still bundle many interactive entries. These are conditional Windows measurements, not competitor comparisons. See performance.

No incompatible public signatures or new obsolete warnings were found against the actual NuGet 0.2.0. The isolated legacy consumer passed source and binary upgrade checks for Docs, Blog and custom templates. Behavioral changes affect previously ignored publication keys, unsafe base URLs and exact diagnostic text; serialized output bytes also change. Follow the 0.2.0 migration guide, including rollback instructions.

0.3.0 targets .NET 10. MDX alone requires Node.js 24.13.0 and explicit restore; its lockfile pins MDX 3.1.1, React 19.2.4 and esbuild 0.25.12. Future 0.x minor releases may change public APIs. See Known limitations for trust boundaries, large-site costs, browser coverage, AOT and AVIF status.

Published Core package providing Markdown-based Docs and Blog output, front matter, custom templates and shared site customization. It remains available as the compatibility baseline for 0.3.0.