# Run trial-balance validation

```powershell
python tests/test_trial_balance.py
python tests/run_parity.py
python tests/build_lab.py
node tests/visual-qa.mjs
python tests/run_acceptance.py
```

Run from this package directory. The Python entry point is `validate_trial_balance(query, groups, accounts, context)`; TypeScript is `validateTrialBalance`. Output lines contain expected raw/normal/debit/credit fields and independent reported differences. Only pass/TRIAL_BALANCE_VALIDATED is a passed projection.

No sibling runtime is needed. The unchanged historical validator is embedded and hash-pinned. Compilation output remains under tests/.tmp-compiled. The lab is self-contained and uses independently authored historical report snapshots; hypothetical reported-value comparisons change copied local variants only.

For current close acceptance the host must require visible p==captured frozen h and retain actual source/chart/genesis/report receipts. A historical pass under p<h is useful for retrospective reporting but cannot certify the later current ledger head.

