[amaybaum]

Documentary corrections to mutable records on main, following the §A.42 evidence landings #1098, #1099 and #1100 (PR-AUDIT-1, #1088). No theorem, definition, proof, import, verdict or frozen record changes, and no evidence is removed.

Owner authorization (amaybaum, Claude Code session chat, 2026-11-04): "Once the evidence landings are complete, authorize committing the approved documentary corrections on a fresh branch and opening one draft PR. Keep the #1009 branch patch unapplied until separately authorized." Later: "The revised ALL-v2.patch scope is approved, including the narrower audit claims and accurately qualified review provenance." Also: "Do not change any scientific theorem, experimental verdict or frozen record." The merge needs separate approval.

Head 6587f87b, three commits on main 82f0e119: e0476bfc (the corrections), a1442f13 (the exact-head review's findings 1–7) and 6587f87b (the delta review's two wording nits). 9 files: 8 modified and 1 added.

Item	File	Change	Owner decision
A	substratum/recursive-substates/README.md	Track A's acceptance is "reported as independently accepted at 7c4e45f7" in #1009 comment 6533749426; reviewer identity and a reviewer-authored verdict are unresolved. A★-1's acceptance at fdeeb2d5 is author-reported (#1011 comment 6546255438), and c811e876 is cited to comment 6543745746	approved as drafted
B1	audits/manuscript/observer-relative-finiteness-20261030/DEPENDENCY-AUDIT.md, synchronized-return row	Narrowed to what control 3 (#872) shows: a binary readout of a countable union of finite cycles whose measured visible matrix is the identity at no positive step. Whether finite-history representability suffices is unresolved (#1015)	"NARROW to exactly what #872 demonstrates"
B2	same file, control 3	#872's construction quoted from comment 6526240419, with its full stated research gate	approved
B3	same file, control 4	#1003: in the ambient one-plaquette SU(2) setting, covariance alone does not uniquely select the Wilson/Yang–Mills action; not a native OI result and silent on continuous internal states or cardinality. #1009: the owner's wording, kept separate. Both broader questions are unresolved (#1015). #1003's README:60–92 quoted at 0efd67a6 (blob 948c7ee8)	"NARROW"; #1009 replacement text approved verbatim
B4	same file, M6	Introduction lines 17, 31, 33 and 67 quoted at #918's fa32b043 (blob afedb761), with the note that lines 17 and 33 differ at #918's head 4028e45d (blob bbf837ff)	"INCLUDE"
C	lean-mathlib/OIBridge/ObservableMeasureReturn.lean (doc comment only), continuous-substates/README.md, equivalence-and-d3-scope.md, observable-recurrence.md	The build citations name the build that checks the landed modules. The README cites main run 41871661965 at 82f0e119 for all three landed modules (ObservableFactorRecurrence, ObservableMeasureReturn, GradientLinkFlatness); its Mathlib bridge Build, Release gate and axiom-report steps all concluded success (job records). The pull_request run 41094894776 for #1009 head a02e197d stays as provenance: GradientLinkFlatness is byte-identical to it, and the Observable modules differ only in doc comments (#1009 commit 798558b7). The a02e197d review (#1009 comment 6529351978) is REVISE, author-posted	approved; "cite the actual successful CI run on the merge commit"
D	oi-qm/reconstruction/composition-access-20261024/SCOPE.md (new)	Which #912 files are on main, with their blobs. #912's branch review/REVIEW.md (blob eceb7060, commit 5b57321a) is an author-posted separate-author ACCEPT of a7f22c38, not an independently authenticated review. MR-1's §2.2 classes #912 as provenance only, but B2 uses its contract (1), theorem and induction proof, and freshness condition, so it is support; MR-1's other uses (:48, N6, T5, §5.6, rows 3.2–3.3, Res, AW-MR3) are listed. #912's controller condition is stronger than B2's, and whether B2's suffices is stated as open (#1088 comment 6636314963). The #885 context is not used	4a "INCLUDE"; MR-1 clarification approved
E	PE-01 pe01_bounded_tests_v1/REPLAY.json	Key unresolved_relative_paths renamed relative_paths; a recipe for re-checking PIN, TH and RP at the run commit 549641460b, since PIN fails at a main head by construction (#1099 delta-review nits N1, N2)	approved as drafted
F	PC-1 physical-coupling-pc1/stage2/REPLAY.json	The four #1022 review records quoted with their verdicts, marked author-posted, reviewer not named on main	approved as drafted

Not in this PR.

• The separate Research (WIP): SUBSTATE-1 — internal sub-states and visible recurrence #1009 branch patch, which stays unapplied; Research (WIP): SUBSTATE-1 — internal sub-states and visible recurrence #1009 is untouched.
• No new review provenance is created: author-posted records are labelled as such.
• Open items stay with PR-AUDIT-1: open pull-request cleanup and evidence-retention dispositions #1088:
  • the "accepted" labels on Research (WIP): SUBSTATE-1 — internal sub-states and visible recurrence #1009 notes;
  • the B2 controller-condition question (PR-AUDIT-1: open pull-request cleanup and evidence-retention dispositions #1088 comment 6636314963);
  • the review's optional finding 8: DEPENDENCY-AUDIT.md:284 still says "Research (WIP): SUBSTATE-1 — internal sub-states and visible recurrence #1009 Lean candidate", although the modules are now on main and built. It is outside the approved items and is left for PR-AUDIT-1: open pull-request cleanup and evidence-retention dispositions #1088.

Checks (local, at e0476bfc, rerun at a1442f13; duplicate, voice and legacy records rerun at 6587f87b).

• Passing: artifact placement, voice, claims, duplicate, staleness, citation roles (self-test and commit), dependency labels, toolchain, the Lean–manuscript census and legacy records (303 intact).
• Both REPLAY.json files parse.
• The patch applied cleanly to 82f0e119.
• No push-triggered workflow matches these paths.
• Closing keywords: none in the commit message or this description.
• Not run locally: the Lean build for the doc-comment change. It runs in CI.

Review.

• Independent exact-head review at e0476bfc: CHANGES REQUIRED, with two blocking findings and six nits.

  • Finding 1: continuous-substates/README.md:273 still cited the a02e197d pull_request run as "exact-head" for GradientLinkFlatness, and called the author-posted review independent.
  • Finding 2: SCOPE.md said B2 uses Composition and access: finite-depth composition contract, logical acceptance circuit and complete-record laws (research record) #912's controller condition.
  • a1442f13 applies both, and nits 3–7.

• Everything else checked out: verbatim quotations, identifiers, line ranges, blobs, frozen records untouched, run 41871661965, and the doc-comment-only Lean diffs.

• CI at e0476bfc (run 41873194586) and at a1442f13 (run 41875257372): 25 checks succeeded and 1 skipped by design, with the Mathlib bridge Build and Release gate steps succeeding.

• Delta review of a1442f13: CONFIRMED WITH NITS, nothing blocking. Its two optional nits are applied in 6587f87b:

  • the doc-comment sentence is scoped to the edit after 82f0e119;
  • "It separates" now reads "The analysis separates".

  6587f87b changes those two sentences in continuous-substates/README.md and nothing else.

Gates.

• Independent exact-head review.
• Green CI at the exact head 6587f87b, including the Mathlib bridge Build and Release gate steps for the .lean doc-comment change.
• Owner approval naming the exact head SHA.
• Merge commit only.

🤖 Generated with Claude Code

https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb

[amaybaum]

Merged. Merge commit d8eda7d84da03ed02407607e6fdf3efca9b38f23, pinned to head 6587f87bae78cde8af4cd1874f0f209736a56412.

Owner approval (amaybaum, Claude Code session chat, 2026-11-04): "I approve a normal merge at this exact SHA, conditional on the head, mergeability and reviewed diff remaining unchanged. After merging, verify the merge commit, tree identity and main CI."

Immediately before the merge:

• The remote head was 6587f87b and the base was main 82f0e119.
• GitHub reported the PR mergeable and clean.
• The diff against main was the reviewed 9 files.
• CI run 41876690537 at the head concluded success: 25 checks succeeded and 1 was skipped by design, including the Mathlib bridge Build and Release gate steps.

After the merge:

• The parents are 82f0e119 and 6587f87b, and the tree equals the head's tree.
• git diff 82f0e119 d8eda7d8 touches exactly the 9 reviewed files: 8 modified and 1 added (composition-access-20261024/SCOPE.md).
• main CI on d8eda7d8 is running and will be recorded with the next landing.

This merge corrects how records on main describe their evidence. No theorem, verdict or frozen record changed, and no evidence was removed. Open items stay with #1088.

---

Generated by Claude Code