Independent source-adjudication packet

101 cases, shuffled and relabeled with opaque IDs. The packet omits implementation results, original descriptive IDs and author validity labels. It is prepared for review; no independent human review has been performed.

Assess each case under its declared snapshot, dialect, profile and payload role. For paired responses, the originating request is supplied. Read the primary sources independently. Suggested assessments: conforms_to_examined_obligations, definite_violation, insufficient_evidence, or source_ambiguous. Record a precise source clause and rationale, including any relevant applicability condition. The task concerns the stated message contract, not prediction of a particular receiver's HTTP status. Absence of a demonstrated violation does not establish complete specification coverage.

Use the blank adjudications.csv. Freeze and hash your completed decisions before seeing implementation outputs or the author's key. Any disagreements should be retained and adjudicated with their evidence. The study author has a separate mapping file, outside this packet, for comparison after review. Do not infer expected decisions from previous versions of the manuscript.

Case provenance: 77 original cases within the context-reference common scope plus all 24 root-flag grid cases. This is a directed source-interpretation review, not a traffic sample. URLs can change; dated source receipts are supplied in the full research artifact.
