ME.10:7 - Conformance Checklist

• A bounded task set names one user System and exact action per row, the actual material or aid, receiving Work or decision, use relation, mandatory or optional criterion, pass and failure observations, window, and stop.
• Human, automated, biological, organizational, computational, and mixed users remain possible; performer, capability, permission, and authority claims use their direct governors.
• In the direct branch, the complete one-result/one-use selection, material loss and direct receiving result remain usable. A later receiving need gets its complete basis once in the owner; an immediate answer needs no new ME.9 row, standalone C.37 account or omission record.
• In the profile branch, every consuming action recovers one complete row with its embedded C.37 claim group, required Method claims, direct governed result, applicable evidence/reliance layers, receiving result, exposure and loss, status boundary, disposition, and return.
• The same branch also recovers one ME.9 cross-use result that names the shared Method source, correspondences, conflicting omissions, edition relations, keep-separate decisions, and changes that reopen several rows. Row colocation is not that result.
• Neither C.37, an action row, nor the ME.9 cross-use result is treated as selecting the support configuration, admitting a Method, publishing material, establishing access, capability, authority, Work, or task success.
• Every candidate configuration is checked against the same task rows, stops, burdens, risks, cross-task effects, and evidence window.
• Observed pass, observed failure, untested, feature-level insufficiency, missing information, and an unresolved use-bounded row are not substituted for one another.
• A sufficient direct route has supported mandatory task coverage, acceptable burden and respected stops; it makes no smallest-or-best claim. A comparative singleton has its declared comparison and selection basis; unresolved comparison preserves candidates, abstention, a split boundary, or a missing priority or test.
• A simple task uses the cheap branch; configuration or recovery testing reuses A.15.8, and non-trivial comparison or selection reuses A.19 rather than copying their kernels.
• Project, user, provider, repository, tool, and automation Systems and their access, assignment, permission, authority, capability, publication, direct-use, Work, and task-result facts remain separate.
• A curated collection has a project or namespace, purpose, current rule, identified membership predicates, edition policy, and return conditions.
• An A.22 selected-structure result has all four discriminator groups plus an actual selection-use basis; otherwise the result remains a proposal and returns the named missing fact.
• Each positive task result rests on independently admitted Work and the direct relations actually observed.
• Provider or AI use, review, feedback, capability, fit, effectiveness, transfer, and authority are claimed only when their own facts obtain.