ARCADIA-Studio/harness/experiments/source-alias-rename-completed-20260912/independent-review.md

Independent archive verification — PASS

2026-09-12, /root/publication_review. No discrepancies found. Read-only verification of archive/original bytes; no archived helper, project code, Cargo or native application was executed. See independent-verification.json and the separate independent_verify.py evaluator.

• Whole ZIP matches archive-result.json: SHA256 2282e4e755bb03ab3ecb079391909f3978297ad307e76306ef7eb5a87a9437cf, 10,755,689 bytes, 77 members, 12,361,849 uncompressed payload bytes.
• Complete outer manifest membership, every payload hash/length and all CRCs pass. All five nested input ZIPs pass exact membership, every input hash/length and CRC: 513/514/514/514/515 members respectively. No duplicate or unsafe paths, executable/DLL/cache/profile payloads, or nested native archive duplicates were found.
• All eight completed command receipts agree with their raw log hashes and summary entries. All 73 original-source readbacks are byte-identical, including the archive builder. External native ZIP 77adda5a… and preserved executable d1ec6305… match their complete SHA256 and byte bindings.
• Raw outcomes match the stated historical scope: attempt6 has 796 tests passed, zero failed, two ignored, plus successful strict Clippy, separate ignored timing and executable build. The clipping negative control fails with the recorded glyph cluster92..93 beneath the scrollbar. The corrected-layout diagnostic run passes 33 focused tests; nested timing passes its one experiment. Editor-patch-check-1 is a successful check, not executed behavior.
• The original candidate f301b8a0…, its documented LFCR finding, and corrected helper 049c9774… inside the completed check snapshot are correctly distinguished. Staged clipping hashes and nested timing summary-to-log binding pass.

Limitations are accurately stated in INDEX.md: this is historical evidence, not acceptance of the current reuse candidate, subsequent evaluator runs, corrected native layout, or release. Native trial payload remains separately bound. Helper sources labeled archive-time readbacks are not asserted to be the exact historical executed versions. Historical uncommitted input snapshots, rather than baseCommit alone, identify each tested source. No new native painting or latency claim follows from this verification.