docs(migration): add living handoff after Phase 6B.2

www/reports/SESSION_HANDOFF.md

@@ -0,0 +1,162 @@
+# AIA Astro Migration — Living Session Handoff
+
+**Last updated:** October 9, 2026 (America/Phoenix)  
+**Repository:** `AIAMarketing/azinstitute4autism.com`  
+**Working branch:** `faithful-astro-migration`  
+**Last completed *implementation* checkpoint:** `6f5ac3c0d233cee5fbbb62be79bc6909e6bec183` (Phase 6B.2)  
+**Next planned milestone:** Phase 6B.3 — Author Archives and Linked Bylines
+
+> This is a short, continually updated orientation document, not a replacement for
+> the code, accepted project instructions, or detailed migration evidence.
+> GitHub branch HEAD may be newer than the implementation checkpoint above
+> because documentation-only commits can follow it. Always verify the branch.
+
+## Start a new session
+
+1. Verify the current GitHub HEAD of `faithful-astro-migration`; inspect recent
+   commits rather than assuming the checkpoint recorded here is HEAD.
+2. Read repository-root `AGENTS.md` (authoritative), `MIGRATION_BRIEF.md`,
+   this handoff, and the relevant recent sections of
+   `www/reports/migration-summary.md`.
+3. For Library architecture, read
+   `www/reports/phase-6a-library-architecture.md` **together with the
+   subsequently accepted Phase 6A ratification in the migration summary**.
+   The ratification supersedes conflicting proposals in the original document.
+4. Inspect the relevant code before drafting the next complete Codex kickoff
+   prompt. Do not presume an implementation is finished because it was planned.
+5. Continue only the next explicitly approved milestone.
+
+## Verified implementation state
+
+- Phase 5C Team reconciliation: `34d5394b3ce765028a27b9cddd9952de3f09bfe7`.
+- Phase 6A Library architecture proposal: `76a035424dc5245443476b168d8c91ea62eed27b`.
+- Phase 6B.1 multilingual metadata search: `b20c253383440598c2d07bf6f852906ac00c42df`.
+- **Phase 6B.2 static Library pagination and redirect definitions:
+  `6f5ac3c0d233cee5fbbb62be79bc6909e6bec183`.**
+  This commit has parent `b20c253383440598c2d07bf6f852906ac00c42df`,
+  message `feat(library): add static pagination and page-one redirects`, and
+  exactly 18 changed files. GitHub verification confirmed all three.
+
+Other earlier milestones and acceptance details are recorded in
+`www/reports/migration-summary.md`. Keep phase checkpoints separate.
+
+### Phase 6B.2 completed behavior
+
+- **65 eligible Library posts:** 46 English, 12 Spanish, seven Arabic.
+  English pagination contains 10/10/10/10/6 cards on five pages; Spanish
+  contains 10/2 on two pages; Arabic has seven on its single Library root.
+- Generated numbered Library routes: `/library/page/2` through
+  `/library/page/5` and `/es/library/page/2`.
+- Existing metadata-only, locale-scoped Library search and Library-only
+  `/search` compatibility survive pagination. Searches use the full locale
+  index, even from numbered pages; clear/reset restores that archive page.
+- Library first-page 301 definitions were added for `/library/page/1`,
+  `/es/library/page/1`, and `/ar/library/page/1`. Isolated nginx redirect
+  tests passed. **These rules have not been activated in the deployed staging
+  or production nginx configuration.**
+- Phase 6B.2 reported 25 Library + 34 SEO + 33 publication + 20 Markdown
+  passing tests, 36 headless Chromium checks, and clean route/link/image/blog
+  audits. No particular screen reader was claimed to have been tested.
+- Normal staging build: **102 HTML routes**, all `noindex,nofollow`, empty
+  sitemap, no hreflang. Indexing-enabled test build: **97 sitemap URLs**,
+  **174 hreflang links over 53 routes**. Restore normal staging build after
+  an indexing-enabled test.
+- Phase 6B.1 separately received manual browser acceptance through the user's
+  built-static-site nginx staging domain `https://aia.web3app.dev`.
+  Do not conflate that evidence with the Phase 6B.2 headless Chromium checks.
+
+## Ratified Library rules to retain
+
+- Use the existing publication-eligible catalog and explicit translation
+  relationships. Ten posts/page; publication date descending, normalized
+  route ascending to break ties. Generate only nonempty numbered pages.
+- Library page 1 is the unnumbered root. Numbered Library pages are
+  self-canonical, indexable in an authorized indexing-enabled build,
+  sitemap-eligible, and **not** numbered-page hreflang equivalents.
+- Retain two accessible English syndicated records in listings and search;
+  preserve their external canonicals, `noindex`, sitemap and hreflang exclusions.
+- Embedded search is metadata-only and client-side with language-specific
+  static JSON. Legacy `/search` accepts previous parameters but initially
+  searches Library articles, **not** the entire website. No external search
+  provider or query tracking.
+- Author archives, once generated, are **initially `noindex,follow` and
+  outside the sitemap**. Only reviewed, indexable author roots may later receive
+  reciprocal hreflang via explicit translation keys; never infer numbered
+  archive equivalents.
+- Do not manufacture missing content or redirect distinct later pagination
+  pages to earlier pages.
+
+## Next: Phase 6B.3 — Author Archives and Linked Bylines
+
+Prepare a **new, self-contained Codex CLI kickoff prompt**, grounded in the
+latest source and the ratified 6A/6B decisions, for this milestone only.
+
+Expected scope, subject to targeted inspection:
+
+- Generate language-specific author archive roots and nonempty numbered pages
+  from exact locale/author references and the shared eligible-post catalog,
+  reusing the ten-item pagination helper rather than manual article lists.
+- Preserve current known author URLs including
+  `/library/author/rula-diab`, `/es/library/author/rula-diab`, and
+  `/ar/library/author/rula-diab`. Current corpus would support five English
+  author pages, two Spanish, and one Arabic. Do not invent English page 6.
+- Link article cards and post bylines to the correct locale author archive.
+  Share exact locale/slug author resolution with BlogPosting Person semantics;
+  fail on missing/ambiguous author records. Keep author and Team membership
+  separate, and do not infer employment changes.
+- Reconcile published author headings and biographies only against traceable,
+  sufficiently current public source material; distinguish potentially stale
+  HubSpot prerenders. Add optional localized `displayName` only if warranted.
+  Substantive Spanish/Arabic biography or navigation text requires human review.
+- Keep author archives `noindex,follow` and outside the sitemap/hreflang graph
+  at this milestone. Keep existing Library-root hreflang unchanged.
+- Plan first-page author alias redirects using the existing redirect registry,
+  respecting the Phase 6B.3/6B.4 boundary and separate host-activation gate.
+- Preserve current search, pagination, publication, staging SEO protections,
+  and static hosting. Include browser/keyboard/RTL/no-JavaScript tests.
+- Do not implement Phase 6C missing articles or deploy hosting rules.
+
+## Outstanding blockers and subsequent milestones
+
+1. **Phase 6C:** Nine missing English Library articles must be reconciled from
+   source before `/library/page/6` and
+   `/library/author/rula-diab/page/6` can be accepted for cutover.
+   Do not generate empty substitutes or misleading redirects.
+2. **Phase 6B.4:** Complete URL, canonical, sitemap, and redirect reconciliation;
+   host activation requires separate approval.
+3. **Phase 6B.5:** Multilingual, RTL, accessibility, and performance acceptance.
+   Spanish/Arabic search and pagination microcopy awaits human language review.
+4. **Later phases:** Other content and language reconciliation (including
+   missing Spanish privacy content), visual/accessibility review, operational
+   integration decisions, and cutover.
+5. Site remains primarily live on HubSpot; production-origin freshness is not
+   proved by cached prerender responses. Stage locally or at
+   `aia.web3app.dev` without activating production forms, analytics,
+   advertising tags, or new PHI flows.
+6. AIA general ABA ages are **18 months through 8 years**; narrower programs
+   have their own eligibility and should not be conflated.
+
+## Working protocol and handoff maintenance
+
+1. Provide a complete, bounded Codex kickoff prompt; Codex implements locally
+   and reports validation plus a structural diff **without staging/committing**.
+2. Review the results and, where needed, perform manual browser acceptance
+   against built static files via the user's Docker/nginx reverse-proxy staging
+   setup. The user runs the serving command from the built output directory;
+   `npm run preview` is not required to duplicate that environment.
+3. After explicit approval, provide a narrow checkpoint commit prompt.
+   Codex commits only reviewed files; **do not automatically push**. The user
+   pushes, and the next session verifies GitHub HEAD.
+4. No production hosting/redirect activation, forms, analytics, search logging,
+   dependency upgrades, fabricated translations, or unrelated modifications
+   without separate authorization.
+5. Preserve the user's untracked root `merge-plan.md` (not stored on GitHub);
+   user/Codex-reported SHA-256:
+   `015db80cdbaf7d68799265d2070db760155c342b432ce6795f17ebc2218c641c`.
+   A remote GitHub write cannot verify this local untracked file.
+6. **Update this handoff at every accepted milestone transition:** refresh
+   last completed implementation checkpoint, next task, acceptance numbers,
+   blockers, and status of deployed-vs-repository redirects. Keep it short;
+   detailed validation belongs in `migration-summary.md`. Preserve the original
+   architecture proposal and dated historical records rather than rewriting
+   decisions retroactively.