Situational applicability specification v0.1 proposed

The SAL specification defines the minimum valid structure of a Better Robots.txt applicability chain.

A chain is invalid if it only says when a capability is relevant. It must also say when the capability is not applicable, what evidence is needed and what an agent must not infer.

Mandatory negative space

The following fields are blocking:

• nonApplicabilityConditions;
• forbiddenInferences;
• requiredEvidence;
• applicabilityAbuseBoundaries.

A generic warning is not enough. The condition must be specific to the domain and chain.

No self-measurement

SAL artifacts may not publish self-assigned scores such as applicabilityFit, agenticUseReadiness, authorityWeight, integrityConfidence or modelCompliance.

The site declares. The auditor measures.