| CC‑A.6.B.1 (Atomicity). | A conforming boundary text SHALL decompose mixed sentences into atomic claims such that each atomic claim belongs to exactly one quadrant L/A/D/E. | Makes L/A/D/E classification unambiguous; prevents contract soup. |
| CC‑A.6.B.2 (Quadrant classification). | Each atomic claim MUST be classified by its own modality and adjudication position, not by its subject-pattern family. When permission wording is present, the single branch in §8.4.1 MUST select the claim's job before assigning L/A/D/E. | Prevents one pattern catalogue from replacing the square's decision. |
| CC‑A.6.B.3 (Form and obtaining constraints). | L-* and A-* claims MUST NOT use RFC deontic keywords as operators. A generic-prescription D-* claim MUST name its exact normative source and applicable rule content without inventing an individual occurrence; an individual-duty D-* claim MUST name its actual bearer and exact separately obtaining U.Commitment; a grant D-* claim MUST satisfy §8.4.1. No claim text makes its relation obtain. Responsibility uses its direct predicate or exact missing governor. An E-* claim MUST name the work, evaluation, or observation that settles it and any evidence used for reliance. | Keeps normative content, individual institution, responsibility, and evaluated results distinct. |
| CC‑A.6.B.4 (Explicit references). | Where a claim depends on another L/A/D/E-classified claim, that dependency MUST be expressed by explicit ID reference rather than restating the other claim in new words. | Prevents paraphrase drift across layers and faces. |
| CC‑A.6.B.5 (E‑claim adjudicability). | Each E-* claim names its exact predicate and object plus the actual work, evaluation, or observation, scope/window, comparison frame, and other conditions required to settle that predicate. It adds an evidence/source-use relation, carrier/schema, viewpoint, and consumer only when the receiving reliance decision depends on that support. | Makes work-effects adjudicable without forcing unrelated carrier apparatus into every result claim. |
| CC‑A.6.B.6 (No gate smuggling). | Operational admissibility predicates MUST NOT appear as L-* laws in the signature layer; they MUST be A-* claims in the mechanism layer. | Preserves substitution and signature stability. |
| CC‑A.6.B.7 (No upward dependencies). | L-* claims MUST NOT reference A-*, D-*, or E-*; A-* and E-* claims MUST NOT reference D-*. | Preserves layering and prevents hidden coupling. |