[amaybaum]

Documentary status repairs 2–7 from the removal-candidate audit under PR-AUDIT-1 (#1088 comment 6638207679). Repair 5 (#912's review record) is the separate PR #1105, so provenance and scientific claims stay distinguishable.

Owner authorization (amaybaum, Claude Code session chat, 2026-11-04): "Status-repairs PR: Prepare the previously approved documentary corrections addressing promotion risks and verification status." "Both PRs require independent review, CI, and separate merge approval." "No deletions, new scientific computations, or changes to frozen records are authorized." The per-repair boundaries quoted below are from the owner's earlier decision on the audit findings.

Head

53a81dfbd820d929276352d0484797abca04d743

• c44fae54 (single parent fb117893, current main): the repairs.
• 4b291c46: the exact-head review of c44fae54 (ACCEPT WITH CHANGES) applied: R1 (all written proofs in continuous-substates listed as unreviewed), R2 (acceptance-label bullet scoped), R3 (Q7 wording also in [U, Q7] items and DESIGN-PULSE references), R4 (PE-01 witness limits from REVIEW.md:148–149, 157–158, 170–179), R5 (two more census notes, RecursiveFiniteFactor → evidence(SUBSTATE-2, #1010): land #1011's 32 branch-only records and modules #1096 and GradientLinkFlatness → evidence(§A.42, landing 3 of 3): land #1009's three SUBSTATE-1 Lean modules #1100, same defect as the PC-1/F2-1 notes), and nits O1–O6 and O8; the delta review of 4b291c46 returned ACCEPT. O7 (MR-1 N5 rests on Research (WIP): SUBSTATE-1 — internal sub-states and visible recurrence #1009's LA-EQUAL) and the stale ANALYSIS.md:3/:56 notes are outside the owner's boundaries and left for follow-up under PR-AUDIT-1: open pull-request cleanup and evidence-retention dispositions #1088.
• 53a81dfb: an ordinary merge of main e0f52232 (evidence(#912): land the review record of composition-access, byte-identical, as author-posted #1105; no file overlap), so the branch is up to date. git diff e0f52232 53a81dfb is byte-identical to the reviewed git diff fb117893 4b291c46.

9 files, no deletion, no frozen record edited, no Lean or code change, no theorem or status label changed.

Repairs and the finding each answers

Repair	Owner boundary	Files
#1009 snapshot	"Add the status block and qualifiers to the mutable citing files. Distinguish reproducible finite controls from unreviewed paper arguments. Do not alter the active #1009 branch or turn author-reported acceptance into independently verified acceptance."	substratum/continuous-substates/README.md (standing block: acceptance labels are author-reported; probes reproducible from main but not run by CI; every written proof in the directory not independently reviewed on main; three kernel-checked modules; §A.46 item 7). Qualifiers: recursive-substates/ANALYSIS.md:176–178; DEPENDENCY-AUDIT.md measured-return row (stale "#1009 Lean candidate" → the kernel-checked ENNReal form and what is missing) and item 2 (solenoid proof is a paper proof). #1009 untouched.
PC-1	"Record the VOID result, build failure, review limitations and owner exception. Correct outdated census provenance without changing theorem statuses or scientific findings."	New physical-coupling-pc1/README.md. Census notes: F2CoherenceControls and PC1Controls ("draft PR #1011/#1022" → landed by #1089, F2-1 records by #1096), RecursiveFiniteFactor (→ #1096) and GradientLinkFlatness (→ #1100); #1011 and #1022 closed unmerged. Statuses unchanged (kernel-only).
MR-1	"State that GAP is provisional, clarify B2's paper-proof and controller-independence limitations, and restrict the broad negative claim to the sources actually examined. Keep frozen records unchanged."	measurement-record-formation/README.md: GAP stands only after a stage 2 not yet run (MR★:876–878; log :745); B2 is a paper proof from #912, whose review record is author-posted, and its controller condition is open (#1088 comment 6636314963); MR★:1353's "in the corpus" falls under C.1-1's sources-read limit.
PE-01	"State the precise witness assumptions and distinguish the demonstrated result from run-only sweeps. If the chapter 18 claim exceeds that support, flag the necessary manuscript correction separately."	pe01_bounded_tests_v1/REPLAY.json: supports.witness_scope with demonstrated and limits (REVIEW.md:21–31, 146–160, 170–179, 247–253) and run_only (the no-passive/no-autonomous-witness statement rests partly on the m = 3 and m = 4 sweeps the reviewer did not enumerate, REVIEW.md:195–196).
O-RET	"Add the status note, correct stale descriptions in mutable records, and state that REVIEW-Q7.md could not be located in the audited repository history. Do not reconstruct the missing review or imply that Phase A establishes physical retention."	New o-ret-routes-20261103/README.md (Phase A scope in the owner's words, not replayed; superseded status lines in PREREG:3–4, IMPLEMENTATION-NOTES:7, DESIGN-PULSE:11, REVIEW.md:148 and :151, DEFINITIONS:133; paper results; REVIEW-Q7.md not located). ANNOTATION.md, which is not a frozen input or line-cited, has its two stale status lines corrected in place. Frozen inputs, the tool commit's notes, the run records and DEFINITIONS.md (line-cited by the frozen preregistration) are untouched.

Chapter 18 (PE-01)

book/ch18-beyond.md:106 (mirror FULL.md:4919): "an embedded observer whose actions reach the hidden state driving a deterministic source can, for some sources, predict its output better than chance." It is existential and names an acting observer, so it is within the demonstrated scope (an active K1 observer using backaction, one realization); the exact-head review confirmed this. No manuscript correction is needed.

Local checks at the head

artifact_placement_check, voice_check, claims_check, duplicate_check, staleness_check, lean_manuscript_census, citation_roles_check.py all exit 0 at 4b291c46; at 53a81dfb, lean_manuscript_census, duplicate_check, staleness_check, artifact_placement_check and citation_roles_check.py exit 0. legacy_records_check.py: 303 records intact at both. REPLAY.json and the census parse as JSON.

Still needed before merge

CI at the exact head 53a81dfb, and separate owner merge approval of that SHA.

🤖 Generated with Claude Code

https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb

[amaybaum]

Merged. Merge commit 150c5b957dfe7051d7b027584c4c42b6c6b69842, pinned to head 53a81dfbd820d929276352d0484797abca04d743.

Owner approval (amaybaum, Claude Code session chat, 2026-11-04): "I approve the normal merge of PR #1106 at exact SHA: 53a81dfbd820d929276352d0484797abca04d743". The owner's stated scope: "Approval is for documentation and verification-status corrections only. It does not establish new scientific results, authorize new computations, or permit removal of research evidence."

Immediately before the merge:

• The remote head was 53a81dfb and the base was main e0f52232; GitHub reported the PR mergeable and clean.
• Exact-head review: ACCEPT WITH CHANGES at c44fae54, applied in 4b291c46; delta review ACCEPT at 4b291c46. 53a81dfb is an ordinary merge of main e0f52232, and git diff e0f52232 53a81dfb is byte-identical to the reviewed git diff fb117893 4b291c46.
• CI run 41893641355 at the head concluded success: 25 checks succeeded and 1 was skipped by design; in Mathlib bridge, Build and Release gate succeeded.

After the merge:

• The parents are e0f52232 and 53a81dfb, and the tree (430a3b97) equals the head's tree.
• git diff e0f52232 150c5b95 changes 9 files (+157/−14) and is byte-identical to the reviewed diff.
• legacy_records_check.py at 150c5b95: 303 records in 75 closed namespaces, all intact.
• main CI on 150c5b95: push run 41897206021 concluded success; 25 checks succeeded and 1 was skipped by design. In the Mathlib bridge job, Build and Release gate both succeeded.

Kept open under #1088: MR-1's frozen N5 rests on #1009's LA-EQUAL, an unreviewed paper proof, which cannot serve as established support until independently validated (the frozen record stays unchanged); and the two outdated statements at recursive-substates/ANALYSIS.md:3 and :56, for a separately reviewed documentary change.

---

Generated by Claude Code