Appendix A — Editorial method and review status
A.1 One document, one source of truth
Breeds and conditions live as structured records (content/encyclopedia_data.py); the authored prose lives beside them (content/encyclopedia_narrative.py). This document, and the Dally Canine Atlas website, and the breed-aware defaults shipped in the app are all rendered from those same records. A breed's weight range is written once and appears identically in all three places.
Consequently, every count in this document is computed at build time. The superseded editions hand-wrote their own coverage claims and drifted: the final one asserted '~185 breeds, ~120 conditions' while actually carrying 398 and 167.
A.2 The record schemas
Breed monograph — the ten authored fields, plus three derived ones:
- Group / classification
- Origin
- Size at withers
- Weight range
- Life expectancy
- Temperament
- Common health issues
- Exercise needs
- Climate & housing suitability
- Owner notes
- Size class — derived from the weight range; selects the normal vital ranges.
- Predisposed conditions — DERIVED by matching the breed's health text against the condition records, giving a real breed-to-condition graph rather than a string of prose. Do not hand-edit it; re-derive it.
- Review status — DRAFT — VET REVIEW PENDING, or the named vet and date of sign-off.
Condition entry:
- Name, aliases, primary body system, and any other systems it also belongs to (so a disease that spans two systems is ONE entry, cross-referenced, not two).
- Urgency — emergency, urgent, or routine. This drives triage on the Atlas and in the app, so it is stated explicitly and confirmed at vet review. It was previously inferred by scanning the description for words like 'critical', which is not a safe way to triage.
- Signs — what an owner would actually notice. Populated for entries written in this edition; empty on older entries, to be filled at review. These are never invented from a description that does not state them.
- Sources — the references the entry was written against.
- Predisposed breeds — the other side of the breed graph, derived the same way.
- Review status, reviewer, and date.
Normal vital ranges are held per size class (from the Part I table) and are what the Dally app uses to seed a new dog's alert thresholds, so a Chihuahua is not monitored against a Great Dane's resting heart rate. Critical thresholds are deliberately NOT derived here — the source table does not give them by size, and they will not be invented; the app keeps its configured defaults until a vet sets them.
A.3 Review workflow
The single authoritative production process for every clinical entry. (It was previously stated twice — once in Part VIII as a Tier-2 plan and once in the closing material — and is consolidated here.)
- Draft monograph produced from source aggregation.
- Adversarial AI check for hallucinated citations, contradictions with peer-reviewed sources, and any inadvertent prescriptive language.
- Structured schema conformity check — every required field populated; every claim carrying a citation or [DRAFT — VET REVIEW PENDING] tag.
- Domain-expert vet review — one licensed vet in the relevant specialty (small-animal medicine for most; cardiology, dermatology, oncology, orthopaedics as needed).
- Editor-in-chief review for encyclopedia-wide voice and cross-reference consistency.
- Publication — with named vet reviewer attribution; version control preserving the review history.
- Sign-off is RECORDED — the reviewing vet's name and the date are written against the entry in the source data. Only then does the entry render as Vet-reviewed in this document, on the Atlas, and in the app; until then it renders as DRAFT — VET REVIEW PENDING.
- Continuous update — flagged for review every 24 months, or sooner if a major clinical guideline is updated.
Nothing that has not completed this workflow may be surfaced to an end user of the Dally app as clinical guidance.