How It Works

Completion is a claim. Evidence qualifies progression.

HubCop is designed to separate the actor doing the work from the decision that allows a consequential milestone to advance.

The proposed architecture in 30 seconds

Work, evidence and the decision to advance remain separate.

1. Actors

People and AI agents do the work

Their “done” is recorded as a claim. A claim cannot qualify itself.

2. Independent sources

Systems other than the actor's own

3. HubCop Evidence Gate

Evidence Recipe: Which independent signals this milestone requires
Policy checks: Identity and context match, independent origin, timing window; same rules for people and AI

PASS → PROGRESS
Next step is permitted
HOLD → AWAITING EVIDENCE
Required signal not yet present; the step waits
ESCALATE → PERSON DECIDES
Signals conflict or the time window expires; routed with reason

Every decision would keep a record of which evidence was used and why.

Specify the source records and independent confirmations that establish this milestone.

Set the identity, context, origin and timing conditions. A signal is not automatically evidence.

Define what pass permits, what hold pauses and where conflicts or timeouts are escalated.

Delivery acceptance. Example data

TODAY’S STATUS QUO

Illustrative status-based flow

WITH HUBCOP · PROPOSED FLOW

AI agent reports “Delivered”
Recorded, cannot qualify itself
Evidence Gate checks 3 sources
Carrier proof, goods receipt, recipient acceptance
VERIFIED: all 3 qualify
Identity, origin and timing match
PROGRESS: invoice proceeds
No manual check needed in this example

The status-based example is not a claim that all AI or workflow systems lack evidence checks.