# DEC-20260106-02 - Data integrity verify

Status: Accepted
Date: 2026-01-06

## Goal

Add automated verification that catches alias/taxonomy drift and encoding-related read issues before merges.

## Intended outcome

- `tools/verify-breed-data.ps1` and `tools/verify-breed-data.sh` return exit code 1 if any of the following are true:
  - Any canonical key in `data/breed_aliases.json` (excluding keys starting with `_`) does not exist as a top-level key
    in `data/breed_taxonomy.json`.
  - Any alias string (case-insensitive, trimmed) appears under more than one canonical key.

- Windows PowerShell reads JSON as UTF-8 explicitly to prevent false failures caused by encoding issues (e.g.,
  `"Vend\u00e9en"` becoming `"Vend\u00c3\u00a9en"`).

## Non-goals

- Fixing historical data problems inside the JSON files as part of this decision.
- Adding a full schema validator for every data file.

## Constraints

- Must run on Windows (PowerShell 7) without requiring a separate runtime.
- Must not rely on network access.

## Acceptance criteria

- The verification scripts fail fast with a clear error message and non-zero exit code.
- The verification scripts pass on a clean repo checkout on Windows and Linux/macOS.

## Rollback plan

- Remove the verification scripts and any workflow hooks that call them.
- Revert the commit(s) that introduced the checks.

## Notes

This is intentionally minimal: it protects core cross-file relationships and prevents silent drift.