[P0][SEO/GEO] Public crawlability, sitemap freshness and indexation verification #98

Objective

Make every published FR/EN editorial article reliably crawlable, canonical and discoverable, without changing the editorial design or the Cloudflare hosting pilot.

Work already started

• PR #96: canonical sitemap generation, localization completeness, hourly cache safety and public route revalidation
• PR #97: avoid the global streamed loading shell on public editorial pages; add HTTP crawler smoke script
• Validate PR CI (lint, typecheck, unit tests); PR fix(seo/P0): canonical and self-healing post sitemaps #96 and fix(seo/P0): preserve human loading skeleton and crawler-readable HTML #97 merged. [ ] Full HTTP smoke and CMS regression remain to be verified independently

Production acceptance

• GET /fr, /en, one FR article and one EN article: HTTP 200; full readable h1, article body and canonical links present in HTML without JS
• robots.txt, sitemap.xml, posts-sitemap.xml, news-sitemap.xml: HTTP 200, valid format and canonical URLs only
• Create / edit / unpublish a localized test article via Payload and the automated import; verify sitemap visibility and cache invalidation; no 404 entries
• Verify IndexNow key file, publication logs, accepted submission status and URLs. Never log secret keys
• Verify Google Search Console property and sitemap metrics via Trigenys SEO Monitor, not an additional GSC service; track exact coverage reasons and crawl timestamps
• Complete issue #90 using the exact post-recrawl GSC error detail; existing News sitemap fix already deployed
• Evaluate and implement preview-safe ISR separately, only after dynamic rendering/crawl checks; do not blindly remove force-dynamic
• Verify latency, navigation loading experience and preview after deploy; retain rollback plan

Boundaries

Existing Vercel production and Cloudflare PR #54 must not be modified as part of the initial P0 PRs. No fabricated indexing data or GEO scores; report measurable observations only.

Post-deployment evidence — 11 Oct 2026

• PRs fix(seo/P0): canonical and self-healing post sitemaps #96 and fix(seo/P0): preserve human loading skeleton and crawler-readable HTML #97 were both merged and Vercel production deployment dpl_G9csRZeTG9AEz4uD1NLskZdhX1dp for commit f07132757e7ac1b667a67a4dcb27a36758f9ad0e is READY.
• Independent public text extraction successfully fetched 6 sampled FR/EN articles (including newly published CITS 2026 and older data-center/finance pieces) with meaningful page content. This does not substitute for a raw-header HTTP status or human-browser navigation test.
• Live posts-sitemap.xml exposes 33 canonical URLs (17 FR and 16 EN); news-sitemap.xml exposes 2 current canonical CITS 2026 URLs; robots.txt names the canonical sitemaps. No off-domain URLs in the collected article sitemap.
• Neon seo_monitor snapshot checked 2026-10-11T13:17:23.506Z: GSC domain property sc-domain:trigenys.com shows 0 errors/0 warnings for News/posts/pages sitemaps; News last downloaded 2026-10-11T03:58:40.412Z. Related SEO: investigate Google Search Console error on news-sitemap.xml #90 is closed with evidence.
• Trigenys Insights GSC page-filter web segment: 82 impressions / 1 click for Oct 2–8 vs 132 impressions / 5 clicks Sep 25–Oct 1, prior to the P0 deployment. FreeLLMAPI EN inspected as indexed; two Cloudflare article URLs inspected as discovered but not indexed. Neither indexation improvement nor impact of PRs is yet proven.
• Draft PR #99 proposes daily public crawler smoke automation.

Still open: canonical/raw HTTP smoke after PR #99, real preview/reader interaction, CMS imports & invalidation, IndexNow logs/config, and measuring future GSC changes. No need to re-open #90 while sitemap errors remain zero.