The acceptance record said every failure's raw output was kept, which
contradicts the record gaps listed right below it (the first a073ccb
derive run and the stage-only diagnostic copies). It now says the
formal test raw output and the earlier review results are kept, no
earlier S evidence is reused for the new S, and the missing original
bytes are stated in the next section. The two ledger references to the
record take its new hash; the static plan check is rewritten.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude Opus 5.5 (Desktop)
34b16de279