Evidence work still open

These are unresolved provenance or checking tasks, not undocumented solver algorithms. The original records remain available and retain their limitations.

Task	What is needed to close it
Published QF_LRA tables	Identify exact Primo commits, manifests, commands, and raw artifacts for Tables 1–2 in the September 2026 paper; verify every aggregate against the ledger. The report currently supplies publication-level results only.
Quantifier reference-less UNSAT answers	Resolve the nine unconfirmed answers in the historical pool-check record, or retain them as explicitly unconfirmed. Do not infer confirmation from a later unrelated gate.
Quantifier export trust	Independently check the conservative-extension conditions of witness exports if proof-like validation is required. The current checker trusts that export contract; see the checking argument.
Historical records with missing metadata	Preserve the original evidence status and name the missing command, revision or artifact. Do not backfill guesses or count them as qualification.

Results-storage migration is separate work. Changing the storage format would not supply missing experiment identities or strengthen an incomplete correctness check.

Publication preparation inspection, 2026-10-02

The ledger at f95ba8a contained 709 historical/qualification entries. Searches for the published QF_LRA aggregate values and likely local paper/tuning artifact directories did not establish a matching experiment identity. A numerical substring hit in an unrelated incremental-UF record was rejected as evidence. The paper report therefore retains its publication-level attribution and makes no raw-artifact reproducibility claim. The owner may have the tuning manifests and measurements elsewhere; no values or commit identities have been guessed.