Data and code for: Modal truncation can reverse predicted crack shielding near the amplification–shielding transition

This unified version 1.3.0 archive provides data, source code, retained states and verification tools for the manuscript “Modal truncation can reverse predicted crack shielding near the amplification–shielding transition”. It preserves the scientific payload from version 1.2.1 and adds 72 states: 42 conventional-modal follow-up states for the hole-edge plate and 30 states for a no-hole centre-cracked plate, comprising 24 reduced predictions saved before six full-order targets. It also contains the saved frequency-translation and retrospective model-disagreement diagnostics, source scripts, fixed policies, modal bases and execution records. The original evidence includes the fixed operating-condition evaluation, 38 completed frequency–threshold pairs, force-history/output-map exchanges, transient checks and 18 full-order endpoint-precision states. Conventional enrichment restores the second sampled shielding direction at twenty modes while magnitude errors remain. The second topology exhibits amplification and shielding at its three fixed frequencies but does not resolve a comparable narrow direction-error interval. The disagreement screen flags four errors among 28 jointly evaluated pairs; ten other pairs lack RF4 and remain unassessed. It is retrospective screening, not an independently validated predictor. The ZIP includes Supplementary Information and standard-library saved-data audits. No physical-joint calibration or universal truncation-failure rule is claimed. Internal preparation-status text is retained as a provenance snapshot from before public DOI assignment; the released record is version 1.3.0, DOI 10.5281/zenodo.23761540. Third-party code retains its documented licence terms.