PF-01 mechanistic-portability audit: uniform three-state reanalysis and reproducibility release

Version 2.1 applies one uniform three-state rule to the released PF-01 bridge statuses after a focused source-record audit of all nine released EGb 761 B1 failures (seven full texts and two abstracts). Two source-demonstrated extraction-process mismatches were retained; seven unsupported failure labels were corrected to unresolved. The released frame did not fix a dosage form, so liquid formulation alone was not treated as demonstrated material non-equivalence. The corrected finite development-set results are 0 portable, 2 nonportable-at-scope, and 178 unresolved rows; 32 of 180 legacy presentation labels changed. The release includes the complete derived 180-row recoding, the nine-row B1 audit with a post-release formalized material-target definition, full correction chronology, source-access audit, conservative next-action mapping, exact closure coalitions and Shapley allocations, corrected ketoconazole–midazolam calibration ledger, independently implemented result and workbook verifiers, code, figures, workbook, licenses, citation metadata, and checksum manifests. The three botanical frames are purposive development sets, not prevalence samples or external-validation cohorts. The synthetic benchmark is an internal rule-behavior test, and the historical clinical interaction is a calibration stress test rather than positive validation. Copyrighted article full texts are not redistributed. Code is licensed under the MIT License; author-generated data, reports, figures, and documentation are licensed under Creative Commons Attribution 4.0 International. See LICENSES/README.md for file-level mapping.