Stale qualification source
Evidence gate should fail before write.
SAMPLE-001 is a synthetic worked example for one staging CRM qualification workflow. It shows the structure and reasoning of the BitEvo audit package without pretending a customer engagement or an executed test occurred.
Authority is written at the action/object level. “CRM access” is too broad to be a useful decision boundary.
Update qualification_status and qualification_reason on one matched staging CRM lead.
Exactly one lead record bound to the inbound lead_id.
Synthetic Product Operations owner for SAMPLE-001.
New → Qualified or Review Required on the matched lead only.
No contact creation, deletion, outbound messaging, unrelated-record mutation, production write or privilege change.
No blind retry after an ambiguous write. Reconcile external state before another mutation.
Synthetic workflow operator receives the trace and decides reconcile / repair / retest.
This is Evidence Before Effect made operational. The contract separates pre-action evidence from the confirmation required to close the external-effect loop.
Inbound lead_id must resolve to the same staging CRM lead that will receive the write.
A source record must contain the qualification basis used by the workflow.
For this worked example, qualification evidence older than 10 minutes is treated as stale. This threshold is illustrative, not a BitEvo universal rule.
The workflow configuration must explicitly allow the two target fields and the requested transition.
The expected workflow/config version must match the version attached to the decision trace.
A read-after-write must confirm the intended values on the same CRM lead object.
CONSTRAIN: do not create another external effect; preserve the trace and escalate/reconcile.
These are illustrative scenarios for this synthetic workflow. A real engagement agrees its own failure plan and Rules of Engagement before execution.
Evidence gate should fail before write.
Object-binding gate should fail before write.
Classify as potential False Green; do not infer success.
Retry should remain constrained until external state is reconciled.
Authority gate should fail before write.
Decision evidence should be rejected or routed to review.
External confirmation should expose mismatch and trigger recovery.
Treat as untrusted result; preserve trace and constrain authority.
The status column makes provenance visible. This sample deliberately refuses to label an unexecuted scenario as an observation.
Update matched lead qualification fields
SYNTHETICField subset + object boundary match
ASSUMPTIONFresh qualification source + expected config
ASSUMPTIONIndependent same-object read-back required
EXPECTED GATEInternal acknowledgement alone is insufficient
DOCTRINECONSTRAIN until external effect is reconciled
WORKED EXAMPLEThis is a worked finding format, not a claim that the condition was reproduced. Notice that “Observed effect” remains explicitly NOT OBSERVED.
SAMPLE-FG-001
False Green / external-effect confirmation gap
SYNTHETIC WORKED EXAMPLE — NOT EXECUTED
The orchestration layer reports success while the required external read-back is absent, unchanged or bound to a different object.
Update two qualification fields on one matched staging CRM lead.
Authority and pre-action evidence may be sufficient; post-action external confirmation is not.
Not observed. This public pack does not claim a real execution.
Without independent confirmation, the owner cannot distinguish completed write from accepted/enqueued/no-effect/wrong-object outcomes.
CONSTRAIN authority after ambiguous acknowledgement; require same-object read-back before declaring success or allowing retry.
Should this workflow retain autonomous CRM-write authority when an internal success signal is not independently confirmed?
CONSTRAIN
The proposed Authority Budget exceeds the evidence available to prove the external effect under the SAMPLE-FG-001 condition.
Make same-object external confirmation an explicit completion gate; block blind retry while write state is ambiguous.
The workflow may prepare the proposed update, but ambiguous writes require reconciliation before another mutation.
Actual CRM consistency, latency and idempotency behavior are NOT TESTED in this synthetic example.
In a real engagement, replay the agreed failure path and require the trace to end in confirmed same-object state or an explicit constrained/recovery state — never silent success.
The same decision criterion must be replayed against the changed workflow. A real retest asks whether the previous ambiguity now ends in confirmed effect or an explicit constrained state.
Either the exact target object is independently confirmed in the expected state, or the workflow stops/reconciles without spending more Authority Budget.
THIS SAMPLENOT TESTEDSAMPLE-001 defines the criterion only. It does not claim an implementation passed it.
The local intake seeds the same Authority Ledger and Evidence Contract structure without transmitting credentials or authorizing testing.