Implementation plan — ordered milestones

Each milestone is committable and runnable on its own.

  1. Schemas + validator + empty dataset — Pydantic models, JSON Schema export, make validate passes on empty-but-well-formed dataset files. ✅ when validator runs and fails correctly on crafted bad data (unit tests).
  2. Dataset authored and validated — all data/*.json populated within §1.2 limits; validator green; flagship etymology chains present and sourced. ✅ make validate.
  3. Core index + explanation engine + tests — graph index, search, deterministic explanations; golden tests for the 10 named entries. ✅ pytest packages/core.
  4. API + smoke tests — all /v0 routes; RFC 7807 errors; /health. ✅ pytest apps/api.
  5. Frontend shell + Lexical Explorer — layout, fonts, search, /lexicon/[id] with explanation + trace. ✅ next build.
  6. Affix Atlas — list, filters, matrix view, detail page with equivalents.
  7. Atom Explorer — list + detail with disclaimer sentence verbatim.
  8. Etymology View — SVG DAG, families, drift, confidence, contested badges.
  9. Docs site + landing — /docs renders docs/*.md; / shows honest counts; /about.
  10. Launch checklist — run docs/LAUNCH_CHECKLIST.md end to end; CI green.