The inspection cycle
Checked on a schedule. Written down every time.
Relgard is not a dashboard you have to remember to look at. It is a cycle that runs whether or not anyone is watching — and it keeps the receipts.
§1The cycle, step by step
The register is read from the source
Each run starts from PA CROP's client register itself — the live list of entities, their types, and their plans — not from a spreadsheet someone exported last quarter. If an entity is on the register, it is in the run. There is no second list to fall off.
Each entity is checked against the Commonwealth
The run checks each entity's Department of State standing through PA CROP's entity monitor — the same machinery PA CROP uses to verify entities at onboarding. A verdict of anything other than active/good standing becomes a flag. A check that cannot complete is recorded as a degraded check and retried next run — it is never quietly treated as a pass.
Every deadline is a countdown, staged
From the entity's type, the run computes its next Pennsylvania annual-report deadline (LLCs: September 30; corporations: June 30; partnerships and nonprofits: December 31) and files it in a stage: clear, within 90 days, within 60, within 30, or overdue. Crossing a stage boundary is an alert event. In live service those alerts go to the entity's owner; in the current shadow run they are logged and sent nowhere, by design.
Every run writes its own audit record
Run started, entities read, checks made, flags raised, failures hit — each is an entry in Relgard's audit log with a timestamp and a run ID. A run that fails writes a failure entry and a non-zero result; the census on this site is computed from those records, not typed in by hand. If the numbers here ever flatter the service, the ledger underneath them will show it.
§2Schedule arithmetic — why daily runs cannot miss a stage
The walk-down stages are 30 days wide at their narrowest (30-day stage, overdue at day 0). The monitor runs daily. A daily run lands inside every 30-day stage at least 29 times before the stage closes — the arithmetic is not close, and that is the point:
| Stage width | Run interval | Runs guaranteed inside the stage |
|---|---|---|
| 90 → 60 days (30 days wide) | 1 day | ≥ 29 |
| 60 → 30 days (30 days wide) | 1 day | ≥ 29 |
| 30 → 0 days (30 days wide) | 1 day | ≥ 29 |
| Grace after due date (~6 months) | 1 day | ≥ 180 overdue flags |
A weekly or monthly check would still land inside these stages — but it would leave no margin for a failed run. Daily runs mean a single failed day is a logged exception, not a missed deadline.
§3What happens when something fails
- The run cannot read the register → the run writes a failure record and stops loud; the census shows the gap.
- One entity's standing check fails → that entity is marked failed for the run, counted in the census, and retried next run. The rest of the run continues.
- A check degrades (the Commonwealth's checker can only say "needs verification") → recorded as degraded, never upgraded to a pass.
- An official mail item sits unreviewed > 24 hours → an exception, surfaced on the daily exceptions view — the queue is not allowed to be quiet.