Proof before offer
The shadow run: sixty days of checking, zero messages sent.
A monitoring service has exactly one job: notice in time. Relgard is proving that job in public before it sells it. For sixty days the monitor runs against PA CROP's live client register exactly as it would for paying members — same roster, same checks, same ledger — except that every alert is written to the log and sent to no one.
§1The terms of the run
| Started | October 28, 2026 — first controlled run completed the same day, against the live register. |
| Roster | PA CROP client entities, read fresh from the register on every run (41 entities on the first run). |
| Checks | PA Department of State standing (via PA CROP's entity monitor) + annual-report deadline proximity, staged at 90 / 60 / 30 days and overdue. |
| Cadence | Daily. Every run writes an audit record; a failed run writes a failure record and a non-zero result. |
| Pass mark | One full alert cycle with zero missed deadlines across the 60 days. |
| Fail consequence | The run restarts from day zero, the failure is recorded here, and the waitlist is told. |
| Client contact | None. Shadow alerts are never sent to PA CROP clients; PA CROP's live service is untouched by this run. |
§2Live census, from the monitor's own records
Known limits, stated plainly: standing checks are name-based against the Commonwealth's checker during the shadow period, and a "needs verification" verdict is logged as a degraded check — never counted as a pass. Definitive per-entity standing keyed to DOS file numbers is a launch requirement recorded in Relgard's assumption ledger, and the run does not complete until the deadline side — the side with dates that can actually be missed — is proven clean.