# Assessor-only conventions These are hand-authored reference guides for the fictional development task set, not scored participant outputs or empirical results. Do not include this folder in a worker or planner's permitted context. A future assessment would follow the protocol's separation between adaptation and post-run assessment; this request only authors the assets. Each task has twelve public requirement IDs and twelve binary criteria, C1–C12. Criterion Cn maps only to public Rn. Award 1 only when all elements stated in that public requirement are satisfied, otherwise 0. The proposed development score is the sum divided by 12; no scores have been assigned. The guides explain acceptable interpretations and source support, not additional obligations. A contradiction elsewhere in a deliverable can make a requirement unsatisfied even if a correct statement also appears. Accept equivalent wording, equivalent arithmetic, and any feasible alternative explicitly permitted by the public packet. There is no required hidden recommendation, schedule, citation style, ordering beyond the specified dependencies, or exact phrase. For word limits, apply the public convention: whitespace-separated words throughout the deliverable, excluding punctuation-only table separators. Source citations such as `[S2:L1-L2]`, separate line citations, or an unambiguous equivalent are acceptable. Numerical answer keys are derived solely from packet facts. Scenarios that allow judgment have acceptable alternatives rather than a single secretly preferred answer. Do not award or deduct marks for production quality, real-world law, unstated best practices, or outside facts. An assessor should flag an ambiguity for a later task version rather than silently resolve it by inventing a requirement. Preserve version 1.0 and record any later changes. The manual reasoning in each guide checks that at least one answer is feasible. It is an asset-authoring check, not an experimental trial, participant completion, or performance claim.