Changelog

The full 5-spec design corpus is now committed and integrated. Spec 3 codified page templates (12 per-cluster templates + cross-cutting rules); Spec 4 shipped the AI-crawler-facing technical files (robots.txt, llms.txt, llms-full.txt, serve.json headers); Spec 5 added the internal-linking audit tool, the 8-action Sprint 3 launch sequence, the Sprint 3+ roadmap, and the 10-metric KPI dashboard.

Operational governance now lives at docs/ops/ — launch-checklist.md, roadmap.md, metrics-log.md. Those files are the ongoing surface; the 5 specs are the stable design foundation.

Audit tool (npm run audit:links) ran once on the 95 shipped pages and found 120 internal-linking violations (silo-matrix gaps + 7 anchor-cannibalization issues). All resolved in a single conformance commit; the audit now exits 0.

Sprint 3 kicks off after this PR merges — 8 off-site launch actions over a 2-week window (GitHub README refresh, awesome-list PRs, MCP directory listings, HuggingFace org, GSC + Bing verification, Show HN submission, Reddit posts, arXiv paper draft) plus 7 content/infra tasks (OG images, benchmarks page, this changelog, arXiv preprint, homepage “What’s new” block, and GSC indexing monitoring).

llmind.org expanded from 11 to 95 pages in one sprint. Full cluster breakdown:

• /compare/* — 5 comparisons (C2PA, Content Credentials, ExifTool, Adobe XMP SDK, NotebookLM) + hub
• /use-cases/* — 9 live use-case pages + hub + 2 stubs (news-media-provenance, mcp-knowledge-base)
• /for/* — 3 audience hubs (developers, enterprise, individuals) + 4 developer archetype pages (agent-builders, dataset-engineers, provenance, dam-integrators)
• /product/* — hub + CLI + web-converter + mcp-server stub
• /mcp/* — hub + Claude Desktop + Cursor guides + 2 stubs (Claude Code, Windsurf)
• /spec/* — 3 LRFS chapters (payload-format, signing-scheme, conformance) added to the existing hub + /ns/1.0/
• /learn/* — 6 new pillars on top of the 4 shipped in Sprint 1 (11 total)
• /glossary/* — 30 term definitions with DefinedTerm schema + hub with DefinedTermSet
• /docs/* — 4 live pages (hub, quickstart, install, cli reference) + 4 stubs
• /blog/* — content collection + 2 seed posts
• Root — /about/, /security/

Shared infrastructure landed: 6 new JSON-LD builders (DefinedTerm, DefinedTermSet, FAQPage, HowTo, SoftwareApplication, ItemList), NAV_ITEMS single-source top-nav constant, GlossaryEntry.astro + StubPage.astro reusable components, DOCS_SUBDOMAIN_ENABLED migration flag stub, and a Playwright E2E harness with assertPageStructure() helper.

83 Playwright structure tests pass; Astro build generates 95 pages.

The first 11 pages of llmind.org shipped: homepage with canonical positioning, /spec/ LRFS stub, /ns/1.0/ stable namespace, /learn/ hub with 4 cornerstone pillars, and /compare/ hub with 2 comparison pages (C2PA, ExifTool).

Site is Astro 6 + TypeScript strict, deployed on Railway behind Cloudflare DNS. Single-source-of-truth site constants in src/lib/site.ts; Organization/WebSite/Article JSON-LD builders in src/lib/schema.ts; BaseLayout.astro wraps every page with SEO meta + Organization schema.

The Spec 1 positioning commitments are all materially visible on the live site — three-layer narrative, canonical one-liner, ruled-out list, three audience paths, LRFS spec framing.