Until this release the skill was organised by topic only. That works until the question is platform-specific, at which point the answer lives in the gap between two files. This release adds a second axis.

Added

astro-seo.md — client:* directives mapped to their INP and LCP costs, server:defer server islands, canonical and Open Graph generation from Astro.site + Astro.url, JSON-LD generated from Content Collections schemas, @astrojs/sitemap including multilingual sitemaps with hreflang, view transitions and their analytics pitfalls, and the SEO-affecting breaking changes in Astro 6 and 7.
edge-seo.md — _redirects design and limits, _headers with X-Robots-Tag and cache strategy, the static-asset vs. Worker-code blind spot, dynamic sitemaps from D1/KV with edge caching, crawler verification by published IP ranges and Web Bot Auth, HTMLRewriter metadata rewriting, and crawl budget.
last_verified frontmatter on every skill file, enforced by CI.
docs/research-notes.md — the verification log behind every factual claim, with sources.
A scheduled workflow that opens a freshness-check issue on the 1st of each month.

Changed

The two skill descriptions were near-identical, so either could fire on any request. English input now selects seo-mastery, Japanese input selects seo-mastery-jp.
README positioning no longer claims freshness is “guaranteed”. What is guaranteed is the process and the date stamp.

Fixed

AI search eligibility was incomplete. The skill said indexability plus snippet eligibility were the only requirements for AI Overviews. Google’s generative AI optimization guide adds a third: the site must be included in Search generative AI features in Search Console.
Article headline has no 110-character limit. Google removed it from the documentation in January 2023. The claim had been presented as a requirement since 1.0.0.
_headers rules accumulate rather than override, joining repeated header names with a comma. The per-crawler X-Robots-Tag example would have shipped one combined header that Google does not document parsing.
_redirects placeholders do match dots inside the path, so /:slug also matches report.pdf. The previous wording made such rules look safe for file URLs.
Google ignores <priority> and <changefreq>; lastmod is only used when consistently accurate. Both are now stated in technical-seo.md.

The full list — including a second review pass that found the dynamic sitemap publishing 404s for hierarchical slugs — is in the CHANGELOG.