FOLD_REPORT.md

pipelines/FOLD_REPORT.md

FOLD REPORT — P1-B + realm-doctrine wave, 2026-07-28

Inputs: 10 audit drafts + 10 fresh-context refutation critics (scratchpad/lens_wave/ and

scratchpad/realm_wave/; the sm64 critic sits at the scratchpad root) plus the 14 binding rulings in

scratchpad/lens_wave/DIRECTOR_ADJUDICATIONS_28.md. Outputs:

scratchpad/lens_wave/REGISTER_V13_FOLD.md (48 new register entries, 46-93) and

scratchpad/lens_wave/CARE_TIERING_V2.md (the rebuilt L-R3 gating deliverable).

Per-unit counts, from the critic closing blocks

Notes on the arithmetic. P1-B killed nothing across 78 findings — the strongest refutation showing

of the program to date, and the ac-traversal, ac-historical and fortnite critics each independently

re-derived their draft's numbers to the digit. The realm wave carried the kills: L-R1's F-01 and

L-R3's F-17 died outright, L-R3's K-1/K-2/K-3 killed the gating table's care column premise and two

of its supporting arguments, L-R2's three strikes are a false-zero control row and two recommendation

clauses, and both corruption kills are MISSING-to-COVERED downgrades the draft had pre-registered

itself. L-R3's 21-minus-4 arithmetic reconciles as 16 clean survivors plus F-01, F-10 and F-18

surviving amended plus F-09 carried contested.

Register selection: 48 of 167 survivors earned a row (HIGH-priority, schema-freeze-relevant or

slice-relevant); the remaining 119 are captured in ten per-lens roll-up lines that name their draft

as the record, so nothing is lost and the register does not gain 119 minor rows.

Could not resolve under the adjudications

Two items, both genuinely outside the 14 rulings and both correctly landing as surfaced rather than

applied. Neither blocks the fold.

• The CVD §6.2 care-tier question (L-R1 F-01 residual, contested item 1). Should Ch 71 Xibalba —

living K'iche' Maya, a living sacred text — sit in §6.2's elevated tier when Ch 47-48 Mesoamerica

does? §6.2 is CVD-enumerated chapter text written for the earthly arc, predating the Realm Road.

Adjudication 8 kills the literal edit; the question itself survives as a decision brief under the

standing protocol and is recorded in entry 72 and in the ledger-corrections block. Nothing is

applied to enrich.py or to the CVD.

• L-R1 F-14's Ch 74 half (contested item 3). The charter's §1.1 records Ch 74 intensity as STANDARD

while the registry reads peak; both divergences are verified real, but this half moves the

premise of a live Josh brief (C-1), and REALM_DESIGN_PROGRAM.md is a director-owned document.

Surfaced, not resolved. The Ch 66 half of the same finding is a clean zero-token repair and folds

as part of entry 72.

Three further items are CARRIED CONTESTED by design rather than unresolved: Ch 72's care_tier

(adjudication 11c routes it to the Wave-2 mint with both sources quoted), zelda Z-15's key-economy

class (re-filed DIVERGENT-BY-DESIGN to CONTESTED with the diegetic seal-object as the default), and

ac-traversal AC-11's reveal adjacency on corrected data (3 waterfall rows, one of which is the Ch-77

reveal site). All three are recorded at their carried state in the register.

Landing checklist for the director

Ordered so nothing writes a file another step is reading, and so the two plan edits that must precede

scheduled work land first.

1. Two PLAN edits, before anything else, because both change a queued work order that is about to

start. Widen window-plan rank 7's scope from "every slice combat surface" to include traversal

surfaces (entry 57). Add a vertical exit test to W-MOVE's verified-by line, or the

traversal_capability_pending tagging fallback (entry 54). Both are one-line edits to

docs/PRE_5090_BUILD_PLAN.md.

2. docs/DESIGN_GAP_REGISTER.md — append REGISTER_V13_FOLD.md verbatim as the new §v1.3 section,

after the existing §v1.2. Then update the §v1.2 board IN PLACE per the fold's claim-flip block:

cells 14a and 24 to CLAIMED-RUN in part, cell 13's boundary pinned, cells 1/2/11/14b gaining their

recorded notes, and the wave status board's P1-B and realm-doctrine lines moving RUNNING to RUN.

Leave every WIDEN-PENDING and OPEN row untouched. Also update the file's Status line to

SYNTHESIS v1.3.

3. docs/proposals/REALM_DESIGN_PROGRAM.md — land CARE_TIERING_V2.md as the charter's per-realm

care tiering section (it is the charter's own named L-R3 output), and in the same commit apply the

three charter corrections the wave verified: §3.3's two-value care_band wording becomes

care_tier per adjudication 10; §1.1's Ch 66 intensity corrected against the registry and the

spine; §1.3's "There is no inside-the-rift canon at all" superseded against the D-2 membrane-pocket

ruling. The Ch 74 intensity half is NOT applied — it goes to the director/Josh brief queue.

4. docs/DOC_MAP.md — add the two new artifacts. Nothing is done until it is listed there.

5. docs/PRE_5090_BUILD_PLAN.md ranked queue — enrol the five zero-token items that need no wave:

the vril-site column repair (entry 46), the LAW 5 scope promotion (entry 47), the citation

predicate on lint_canon_graph.py (entry 59), the teaching-order producer/consumer naming

convention on the already-armed gate (entry 64), and the zone-catalog realm-row blocker (entry 70,

which must be flagged as the first Wave-0 action because the scheduled populate currently no-ops

silently).

6. Wave-0 vocabulary reconciliation, as one commit before any realm tooth is written: terrain,

vril_density and the newly-found ambient_threat five-versus-four disagreement, plus the V seed

set and its out-of-vocabulary token (entry 76).

7. The canon-bearing citation repair (entry 68) — director-level apply, enumerated by beat id, fresh

critic, full run_gates.py suite. Never re-derive "all HL_0003 hits."

8. Any registry edit in the above (entries 46, 66, 72, 80, 82, 91) refreshes

docs/fidelity_baseline.json via --emit-baseline on registry_fidelity.py in the SAME commit,

or the gates go red.

9. The ledger 28g entry cites the fold's "Corrections to the ledger 28f headlines" block by name —

F-01 killed, F-10 scope-cut to roughly 10-11 — rather than rewriting 28f.

10. Two pre-handoff cleanups that are mechanical and should not consume a critic: re-anchor the

PRE_5090_BUILD_PLAN.md line refs in the zelda draft (CRLF drift, running about 24 low) and the

four wrong REALMARC section numbers plus five body-prose line cites in the register-variety draft.

Both drafts are written for direct pickup by a schema author who will follow every pointer.

Generated by harness/site/structure_site.py — the URL path is the repo path. review root