SCO Judgment Layer Compatible
Data platforms, models, agents and workflow systems each handle part of an organization’s work. They do not usually carry a durable representation of what an organization currently believes, how uncertain it is, what alternatives it has entertained, why a judgment changed and whether prior judgments were borne out.
When two systems exchange a judgment, that representation can survive the exchange or be lost in it. This profile lets a third party check which of the two happened. Whether the judgment is sound is a separate question, which the companion benchmark measures.
Four independent classes
The classes are independent of one another. A read-only dashboard can be a Consumer without being a Producer, and an attestation service can be Attested without parsing a single view.
Assessed so far
The first entry is the maintainers’ own reference implementation. Entries are append-only, so a system that later regresses keeps its entry, with an annotation.
The compatibility validator
The validator is deterministic and runs offline, with no network access, no model calls and no account. It is available on request from the project, so that a party other than the one making a compatibility claim can check it.
What a statement does not establish
A statement does not establish that the judgments carried are true or that the reasoning behind them is sound, which the companion benchmark measures. It does not show that a recorded rationale is the reason the system reached its output, that everything material was recorded, or that the artifact reflects the model’s internal state.
At the level of the system, no security or privacy property is assessed. A listing is not an endorsement, and a statement describes only the version assessed on the date assessed.
A candidate profile
SCO-JLC 1.0-draft is a candidate profile, open for anyone to read. Assessment is run with the validator and re-verified by the maintainers before listing, and other implementers are welcome to submit a system.