Evidence-scoped conformance

A claim is only as strong as the proof behind it.

Veklom separates source presence, tests, runtime observation and real consequence evidence. The site does not convert a configured feature or passing health check into a stronger claim than it earned.

01

Authority

No consequence beyond granted authority.

02

Evidence

No truth claim beyond observable evidence.

03

Agency

No residual execution authority after termination.

04

Monotonicity

Derived authority may preserve or narrow scope; it does not silently widen.

05

Uncertainty

Unknown outcomes stay unknown until reconciliation.

06

Fail closed

Missing or invalid authority denies before consequence.

Proof classes

One vocabulary for what is actually known.

This is the language the public product should use everywhere: source, tests, live runtime, verified consequence, or unverified. Nothing gets promoted because it sounds better in marketing.

01SOURCE_OBSERVEDCanonical source contains the behavior. Deployment is not implied.
02TEST_VERIFIEDAn executable falsifier passed in the stated environment.
03RUNTIME_OBSERVEDA declared service responded in the tested runtime profile.
04CONSEQUENCE_VERIFIEDA real consequence and authoritative post-state/evidence were observed.
05UNVERIFIEDRequired evidence has not yet been produced.
Deployment truth

A healthy plane does not certify the whole system.

BYOS Runtime, LockerPhycer, CAPPO, cAPI, Gnomledger/PGL, VLink and Guardian have separate responsibilities. The live proof page reports individual runtime observations rather than collapsing them into one global green badge.

Regulatory boundary

Evaluation is not certification.

Veklom can be evaluated against privacy, security and AI governance frameworks, but the public product should not claim legal compliance, regulatory approval or universal alignment without a scoped assessment and deployment-specific evidence.