🔧 Auto-commit from sysupdate on 2026-06-03 21:09

2026-06-03 21:09:27 +02:00
parent 29b3d43970
commit bc53c94742
14 changed files with 532 additions and 2 deletions
@@ -81,6 +81,7 @@ mas "Super Agent", id: 1568262835
 mas "Table Tool", id: 1122008420
 mas "TestFlight", id: 899247664
 mas "Vinegar", id: 1591303229
 mas "Xcode", id: 497799835
 mas "xSearch", id: 1579902068
 npm "@anthropic-ai/claude-code"
 npm "corepack"
@@ -765,4 +765,28 @@ The Jurist's six-phase authorization map governs sequencing. Cross-cutting succe
 **Audit update (2026-05-28):** four-pass pre-build audit completed (`pre-build-audit-2026-05-28.md`). Findings (a)/(b)/(c)/(d) were re-tested against substrate; A1's persistence-finding (no schema for `means_of_knowing`; only numeric value of `EarnedConfidence` persisted) corrected; B1's "prior-only constraint" borrow ruled redundant (BMF enforces by construction); B1's `_cap_links_per_unit` borrow validated as 1–3 lines; the parent amendment's items 9 (numeric confidence in recall response) and 10 (epistemic state in health) found NOT shipped, reshaping A1's scope. Co-author branch deferred. Audit document is part of the Monday package.
-**Awaiting:** steward review of the audit; jurist review of A1/A2/B1; co-author engineering review of B1/C1/C2/D1 (Seb on resumption from Peter block 2026-06-01+; steward + executor continued joint work during).
+---
 ## PENDING-25 — Translated renderings in the ARC corpus
 **Date:** 2026-06-03
 **Tag:** [PROPOSAL]
 **Summary:** A doctrine + frontmatter convention for translated/multilingual works — ARC holds *works*, of which languages are non-canonical *renderings*; the work (not any rendering) holds the corpus slot.
 **Rationale:** Lune will translate *After the Reply* into French — the first translation in the corpus, and a first instance of intergenerational stewardship in ARC. The corpus had no settled mechanism (only a tentative `related_texts` line in frontmatter-spec §8, and one accidental Context-C legacy piece). Jurist's Opinion (2026-06-03) ruled: Option B *reframed ontologically* — neither rendering is canonical, both express one work; Context C legitimate-but-gated; `translator:` a **required** field (intergenerational dimension carried structurally); amend frontmatter-spec (no new companion).
 **Options:** Work-level corpus-slot representation — W1 (`primary_rendering` flag; rejected, re-smuggles privilege), W2 (work-stub object; faithful but heavy), **W3 (group by `work:` at build; AUTHORIZED)**. Sub-questions resolved 2026-06-03: routing = asymmetric (root slug stays, additional renderings `/<lang>/<slug>`); rendering-date = each rendering's own publication date; Context-C-placement = frontmatter-spec §8 + §I.j pointer (settled). **Only toggle affordance remains open** (deferred to implementation, to be rendered not specified blind).
 **Recommendation:** Adopt the jurist's rulings; implement via W3. Full proposal: `docs/AldineXXI-Codex/drafts/translated-renderings-proposal-2026-06-03.md` (jurist brief at `multilingual-translation-handling-jurist-brief-2026-06-03.md`).
 **Files affected (on implementation):** `frontmatter-specification.md` (§2 new subsection, §4 recipe, §8 ruling); `site.hs` (listing dedupe by `work:` + representative-selection); piece template (rendering toggle); routing rule. None touched yet.
 **Status:** AUTHORIZED-with-conditions by steward+jurist 2026-06-03; the four conditions are addressed in the proposal document (date resolved, `work:` omission corrected, §7.3 settled, routing decided). Formal REVIEWED entry is the steward's to file. Implementation **split along its natural seam** (executor analysis 2026-06-03, steward-accepted): (a) the `work:` listing-dedupe awareness folds into **F-Phase 2** (the Hakyll listing generator — same `site.hs` listing surface; build it work-aware from the start so the generator is never reopened; dormant until a `work:` field exists, like §I.k); (b) the rest — frontmatter-spec amendments, toggle template, routing, heteronym decompose — is its own focused session **after Stage G** (stable typology), with the toggle *rendered for the steward*, not specified blind. **Decoupled from the Vignette** (independent — touches neither listing nor rendering). **Test vehicle = the Matthieu Taliesin heteronym refactor** (steward 2026-06-03): decompose the trilingual piece to en/fr/es renderings as the proving ground for the toggle+`work:` machinery (first application of the §5 Context-C gate, not a reversal of jurist's no-force-decompose ruling). Also incoming as renderings: steward's bio/CV + About pages (little-by-little). **Readiness target: late August 2026**, ahead of Lune's *Après la réponse* (expected late Aug/Sept). Independent of §I.k.
 **Awaiting:** steward review of the audit; jurist review of A1/A2/B1; co-author engineering review of B1/C1/C2/D1 (Seb on resumption from Peter block 2026-06-01+; steward + executor continued joint work during).
 ## PENDING-26 — `.whisper` retirement: §XIV name-lock + dedicated migration pass
 **Date:** 2026-06-03
 **Tag:** [HARDENING]
 **Summary:** Retire the unsanctioned `.whisper` class (apparatus-spec 462–464; AldineXXI Selector Index 2801) via a dedicated, gated migration pass — after steward/jurist lock the target apparatus class names per §XIV.
 **Rationale:** Stage N's deep read corrected the scope: the closing plan recorded "27 template usages"; ground truth is **~156 line-usages — 11 templates (~14) plus 100+ content markdown files (~140)**. This is a corpus-wide migration, not a residual deletion. `.whisper` conflates two semantic roles that must not land in one class: **(a) apparatus prose** (footer text, list metadata) and **(b) list-context qualifier glyphs** (`✦`/`⟐` — the same marks that render via `.chamber-mark`/`.encounter-mark` on post pages). Stage M left a non-regressing holding value (`_utilities.scss:32`, `var(--gray)`); the §VII.e batch (commit `20019ba`) did not touch it. The jurist's Stage-N review requires this carry to be a named PENDING item with commit-gate, not a verbal deferral.
 **Name candidates (surfaced from the specs, for the lock):**
 - Role (a) apparatus prose → **`.apparatus-meta`** (AldineXXI Selector Index 2801 "migrate to `.apparatus-meta` or equivalent").
 - Role (b) list-context glyphs → **`.register-mark-inline`** (apparatus-spec 464) **or unification with the existing `.chamber-mark`/`.encounter-mark`** (arguably the same mark in a quieter context — `_post.scss` historical note: ⟐ rendered via `.chamber-mark` on pages, `.whisper` in lists; one mark, one class, context handles register).
 **Options:** (1) Jurist rules on the names in a focused review; executor then runs the migration as its own pass — scripted edit across templates + content, render-verified per type per viewport, sequenced **before** the full re-partition. (2) Steward locks names directly in REVIEWED-26 annotation; same execution path.
 **Recommendation:** Option 1 — the role-(b) choice (new class vs unification) is a genuine taxonomy question the jurist should rule on; unification is the leaner answer but touches the qualifier-mark grammar (§V).
 **Files affected:** ~11 templates, 100+ content files, `_utilities.scss` + `_layout.scss` (the holding rules), apparatus-spec + Selector Index cross-references.
 **Awaiting:** Steward authorization + jurist name-lock. Until then the holding value stands; no migration work proceeds.
@@ -1,13 +1,16 @@
 # Claude Code Memory
 ## Standing preferences
 - [Typography palace — query via CLI](reference-typography-palace-cli.md) — the type masters (Bringhurst/Tschichold/Hochuli/Lupton/Sousa, ~20-21 sources) live in a dedicated MemPalace at `~/.mempalace/palace-chamber-typography`; query with `mempalace --palace <that> search "…"` (CLI, NOT the MCP tools which point at palace-memory). For ARC typography work — don't re-ask where it lives.
 - [Fowler's rules for quotation](feedback_fowlers_rules_quotation.md) — single quotes primary, double for nested, logical British punctuation order. Apply across all writing.
 - [MemPalace KG object 128-char cap](feedback-mempalace-kg-object-128-char-cap.md) — `kg_add` `object` hard-caps at 128 chars; write KG objects as short keyword phrases on the FIRST pass, detail goes in the drawer. Recurs at every /wrap-up §5 — stop re-deriving it.
 - [Rank on fields you actually write](feedback-rank-on-fields-you-actually-write.md) — a consumer that ranks by an evaluative field (importance/weight) nothing populates silently degrades to trivial order while claiming to rank; verify scoring fields end-to-end, prefer signals already captured (recency). For BMF/CapableMind/studium-engine tool-building.
 - [Trust prior pass frame](feedback-trust-prior-pass-frame.md) — When extending a prior verification, re-run it at the scope of the extension. Frame-inheritance from a deep-read is contamination shape: the prior pass tested what it tested; your extension claims things it did not test. Caught 2026-05-28 → four-pass audit. Verify-before-compose at draft time, not build time.
 - [Verify each post type after changes](feedback-verify-each-post-type-after-changes.md) — ARC: after any shared CSS/template/composition change, check EVERY content type on BOTH desktop and phone, not a sample on one viewport. Lesson from §II.d gate verified phone-only → desktop appearance surfaced 2 sessions late (no regression; just unseen). Per-type×per-viewport extension of [[trust-prior-pass-frame]]; belongs in the SCSS verification ladder.
 - [Reading Compass is SVG, not type](feedback-reading-compass-is-svg-not-type.md) — ARC Compass sigils are SVG `<symbol>` artwork, NOT Plex Sans / nav typeface / Pattern 3. Recurring conflation; steward corrected 2026-06-03. Plex-Sans nav *text* = breadcrumb + version-stamp + post-type only.
 - [MemPalace is an unaffiliated stopgap](mempalace-is-unaffiliated-stopgap.md) — MemPalace is third-party; steward/Seb build BMF/CapableMind (hoped to replace it). Don't conflate them; MemPalace PRs await MemPalace's maintainers, NOT Seb.
 - [Skill-harvest register](skill-harvest-register.md) — **canonical home for proposed skills** (create/patch/retire) awaiting steward authorization; the governed analog of PENDING.md for our own tooling. `/wrap-up` §1.6 proposes here; steward authorizes; `/wake-up` surfaces open proposals. Check at wake; append at wrap. Open now: `/spec-code-audit`, `bmf-diagnose`, `/arc-typeset`, `/pre-build-audit`, `/vignette` (deferred to Phase 2), Symmetria `trust-prior-pass-frame`.
 - [Plane coordination workflow](reference-plane-coordination-workflow.md) — **NOW CENTRAL to the steward↔Seb workflow** (est. 2026-06-02): `app.plane.so/capablemind`, BMF project, thin effort-tasks `[BM #N]` over canonical GH issues (effort-only, no technical detail; labels/priority mirror GH; issue attached as link). Executor keeps task state current as the 6-hour relay. ⚠ BBF = BetterBridge *product*, NOT the board. Plane MCP user-scoped/OAuth; load tools via ToolSearch `select:mcp__plane__*`. `create_work_item` needs `description_html`.
 ## Canonical Workstream Trackers
 *Read the tracker for any active workstream at /wake-up before composing the briefing. Append substantive moves at /wrap-up. Per `feedback-canonical-workstream-tracker-discipline.md`.*
@@ -18,6 +21,18 @@
 - [L1 reliability](project-L1-reliability.md) — canonical workstream tracker established 2026-05-28 (Symmetria-pulse decision). Current state at top + Active umbrella (PENDING-24 six sub-items) + four-findings table + sources of truth + decisions made/deferred + skill-harvest candidates + chronological log 2026-03-21→. Active goal: **working proposition for Seb by Monday 2026-06-01** (steward 2026-05-28, "bring him relief"); deadline shifted to mid-July. Read tracker at /wake-up before composing L1 portion of any briefing.
 ## Active Session
 - [Session 2026-06-03 (evening) — ARC Stage N: prune + §VII.e conformance + the threshold, PUBLISHED](session-2026-06-03-evening-arc-stage-n-conformance-threshold.md) — **Stage N nearly closed in one evening; 6 commits, pushed + deployed + live-verified** (`e0f35db`→`3f89564`). (1) **Dead-glyph prune byte-proven** (5 fields incl. a 5th the wrap missed; 2 dead templates; `currentSymbol`; `needsGlyph`→`rendersAsPost`; `conditionalField`; `diff -r` = 0 twice; jurist ACCEPTED). (2) **§VII.e batch, measured first** (separator gray@0.5 = 1.84:1 FAILS 3:1; breadcrumb category gray = 3.98:1 FAILS the 4.5 TEXT floor → grey off the table on a11y): apparatus text → fg recede-by-form; opacity attenuations dropped (+4 list-context duplicates caught at the **compiled-CSS grep gate**); **breadcrumb componentized** (steward's eye caught sequence-index + about breadcrumbs silently in Garamond — one `.breadcrumb` rule, 3 templates); category fg + xs (**A1 settled by the masters: Bringhurst running-heads + RUTTER from the Chamber library** — `converted_texts/collections/typography/web-typography-rutter.md` — who VALIDATES Plex-Sans-as-functional-face and rules size the lever, weight out; **nav-face reconsideration closed, no proposal**). (3) **Frontispiece threshold**: epigraph + attribution → fg (steward chose on render); **`font-variant: small-caps` drift FULLY retired** → canonical EB Garamond SC mixin (cite + title + title-return; silent late font-family override caught at the compiled gate); title = one even-SC lapidary line; phone-centering fix (mobile blanket `p` rule). "It is much stronger." (4) **PENDING-26 filed** (`.whisper`: ~156 usages = 6× the plan's figure; name candidates `.apparatus-meta` + `.register-mark-inline`-or-unify; jurist to lock). **NEW CARRY:** frontispiece **compass reveal mechanism never implemented** (dark mode exposed it; steward: own pass, to the level of everything else). New memory: `feedback-reading-compass-is-svg-not-type`. **Pulling thread: CLOSE STAGE N** — systematic dead-selector sweep + §I.h rhythm token; **Option B Compass-§VI-as-data** as the paired build; then re-partition → Stage G → vignette. **Literal Q:** REVIEWED entries for PENDING-25 + -26 filed? and did the jurist pick `.register-mark-inline` or unification? ARC `main` @ `3f89564` clean/pushed/deployed; `~/dotfiles/PENDING.md` carries -25 AND -26 uncommitted. Skill harvest: reinforce verification-ladder with compiled-selector grep + WCAG-contrast calculator.
 ## Archived (2026-06-03 morning — §I.k + PENDING-25 + Phase-2 listing generator)
 - [Session 2026-06-03 — ARC §I.k French spacing + translation-renderings doctrine (PENDING-25) + Stage-F Phase-2 listing generator SHIPPED](session-2026-06-03-arc-ik-french-spacing-translation-doctrine-phase2-listing-generator.md) — **A long ARC build day; three threads all completed, the last deployed live.** (1) **§I.k Build-time Typographic Enforcement** (`02373fd`): `frenchSpacing` Pandoc AST transform, uniform U+202F before `; : ! ? »`/after `«`, gated on `lang: fr`, upgrade-not-insert (URLs safe), idempotent; **jurist ruled uniform-U+202F** over the Lexique's fine/mot distinction (web-transposition; companion §1 stands); verified on synthetic corpus + English gate; dormant (no `lang: fr` piece yet). (2) **Translated-renderings doctrine → PENDING-25 [PROPOSAL] AUTHORIZED** (jurist brief → Opinion: ARC holds *works*, languages are non-canonical *renderings*, the work holds the corpus slot; `translator:` required; Context C gated; amend frontmatter-spec, no new companion; W3 = group by `work:` at build; 4 conditions addressed; impl deferred — listing-half folds into Phase 2, rest after Stage G, decoupled from vignette; **heteronym refactor = the toggle's proving ground**; triggered by **Lune translating *After the Reply* → French, ~late Aug/Sept**). (3) **Stage-F Phase 2 SHIPPED LIVE** (`4303df0`+`6851367`+`41fc030`, deployed + verified at animalrationiscapax.org): 11 per-type `create` blocks → **one generator** over new `data/listings.yml` (presentation facts; validated vs `content-types.yml` enumeration → unknown class = build error), **byte-IDENTICAL** listings; dead triptych/glyph JS retired (`<head>` load + `assets/js/glyphs/` ~100 files); steward's two probing Qs caught stale assumptions (**`currentSymbol` is DEAD — SVG `<symbol>` per-slug is the rendered sigil since 2026-05-26**; SVG stays out of jurist-pending typology §2). Wrote **`arc-horizon-2026-06-03.md`** (one-place remaining-ARC view: #1 done → Stage N → re-partition → Stage G → vignette). **Returns:** the steward's questions were the engine of quality — verify-current-state/render-and-LOOK, don't reason from the remembered model; governance loop held (surfaced fine-vs-mots to jurist, drafted doctrine as brief→PROPOSAL); proportion held ("do it once well" = two right-sized once's, not a Compass-data mega-change). **Pulling thread: ARC Stage N** (spec-conformance + dead-code prune), now loaded with Phase-2 dead-glyph fallout (rename live-but-misnamed `needsGlyph`; delete dead `renderGlyph.html`/`partials/triptych.html` + `has_glyph`/`glyph_id` field defs + vestigial `currentSymbol`; SVG-vs-unicode spec note) — with the bounded **Compass-§VI-as-data follow-on** (`listings.yml`+specials → nav rows; closes the (slug,title) overlap Phase 2 opened) as the satisfying adjacent re-entry. **Literal Q:** subtractive Stage-N dead-glyph cleanup first, or the Compass-data follow-on? + has the steward filed PENDING-25's REVIEWED entry? **CapableMind deferred entirely** (steward); housekeeping remainder parked. ARC `main` @ `41fc030` clean/pushed/deployed; `~/dotfiles/PENDING.md` carries PENDING-25 uncommitted (steward's repo). Skill harvest: extend the SCSS verification-ladder candidate into a general **build-output byte-equivalence gate** (proved 3× now: SCSS, site.hs generator, §I.k transform).
 ## Archived (2026-06-02 afternoon/evening — Plane.so coordination layer + MemPalace cold-start fix)
 - [Session 2026-06-02 (afternoon/evening) — Plane.so coordination layer stood up + MemPalace recurring-drop ROOT-CAUSED & FIXED (#1495 cold-start warmup)](session-2026-06-02-plane-setup-mempalace-coldstart-fix.md) — **CapableMind housekeeping pivot; two arcs, both shipped.** (1) **Plane.so bridge** (Seb's ask — make the 6h gap a relay): added official **Plane MCP** (OAuth, user-scope, no secret); model = GH issues canonical/technical, **Plane task = thin effort overlay** (`[BM #N]` tag, effort-line only, GH link, labels/priority mirror GH; coupled-not-mirrored). Home = **BMF** project (cleared 7 demo cards, seeded 5 thin tasks #165/#167/#170/#168/#171 + 5 GH-mirrored labels). **BBF=BetterBridge product, NOT the board** (steward corrected my misread→read the spec). Recorded central in `reference-plane-coordination-workflow.md`. Steward: shape+assignees "wait & see"; grow-as-we-go. Gotcha: `create_work_item` needs `description_html`. (2) **MemPalace silent-drop fixed** (was "root cause unknown" for weeks): steward's auto-write-keepalive hypothesis DISPROVEN (never disabled; separate process). Real cause = upstream **#1495** — embedder cold-load on first chromadb-write (esp. `diary_write` at wrap) exceeds MCP timeout → -32000, no trace. Fix `cf96e68` already in tree (`local/bge-m3-on-3.3.6`, 0 behind) but **opt-in**; steward flagged non-standard bge-m3/MPS engine → VERIFIED warmup is embedder-agnostic (loads bge-m3/MPS, cached, wired in main() L2845). APPLIED `MEMPALACE_EAGER_WARMUP=1`+`MEMPALACE_LOG_FILE`+`MEMPALACE_MCP_IDLE_HOURS=0` in settings.json env; **needs restart to activate**. Patched `/wrap-up §4.0` liveness-ping (steward-auth). **Returns:** steward corrected me twice (BBF; non-standard engine) — verify-before-assert through-line; time-discipline held (upstream-issue check WAS the win). **Pulling thread: VERIFY the MemPalace fix holds after restart** (mcp.log shows warmup, first diary_write clean — applied-but-unverified). **Literal Q:** did the warmup take for bge-m3/MPS, or escalate env placement? Housekeeping remainder (untracked CM-AI files, PENDING bloat, BMF parked, L1 tracker stale, Seb gate-Q) untouched.
 ## Archived (2026-06-02 morning — ARC craft day; four deploys, all verified live)
 - [Session 2026-06-02 — ARC verse typography + Pandoc F1 + icosahedron mark + favicons (FOUR deploys, all verified live)](session-2026-06-02-arc-verse-pandoc-f1-icosahedron-mark-favicons.md) — **A long ARC craft day; every thread shipped, nothing mid-flight.** (1) **§I.h.b/c Verse typography** (`5318ae1`): read masters from the **`palace-chamber-typography`** MemPalace (CLI access saved as `reference-typography-palace-cli.md` — answers last wrap's open Q); steward caught that 'Time is not a river' is HIS poem (primary verse) not a quotation → **split §I.h.b Verse Quotation (indent) + §I.h.c Primary Verse (no indent)**; `::: verse`+`| ` line blocks, `Ext_line_blocks` enabled, shared `.verse` SCSS; re-authored the meditation + set the **real Berger poem** into Vespers. (2) **Pandoc utilization audit** (deep): **F1 SHIPPED** (`8549ead`) — zero real math in corpus, so retired `tex_math_dollars` + the 40-line `$`-escaping hack + MathJax (−70 lines); **F5 HELD** (`writerSectionDivs` conflicts with §II.d marginalia grid; needs `display:contents` a11y resolution first); **F4 queued** = `§I.k` build-time typographic enforcement (French-NBSP filter as Haskell `pandocCompilerWithTransform`, spec-FIRST); artifacts `_audits/pandoc-utilization-audit-2026-06-02.md` + `pandoc-hakyll-toolset.md` (indexed in CLAUDE.md, `956542c`). (3) **Icosahedron site mark** (`4ef26b0`): redrawn the right way through a full reference lineage (Garamond→manicule→Jamnitzer→Roman d20→engraving) → a **solid shaded faceted die**, inline mode-aware `var(--accent)`, defined once in `templates/site-mark.html`, og raster re-rendered (1.4MB→45KB). (4) **Favicons** (`7ec44aa`): regenerated from the new die + fixed 3 pre-existing bugs (empty manifest name, broken icon paths, stray `#5bbad5` mask color in BOTH head templates). **Returns:** render-the-artifact-and-LOOK before asserting (caught my own feTurbulence/table errors); steward corrected my flattened verse distinction + my under-committed "restrained" mark rec (the gesture wanted boldness). **ARC fact to keep: `<head>` is duplicated across default.html + frontispiece.html — any head change touches BOTH.** ARC `main` @ `7ec44aa`, clean, pushed+deployed. **Pulling thread = a CHOICE for tomorrow (threads all completed): ARC F4/§I.k (French-NBSP, spec-first) vs the parked L1/Seb proposition.** Literal Q: which workstream tomorrow?
 ## Archived (2026-06-01 — Stage M published live; the verse-spec gap today resolved)
 - [Session 2026-06-01 — ARC Stage M complete + PUBLISHED LIVE; .post-content + .callout purges; verse-spec gap surfaced](session-2026-06-01-arc-stage-m-published-postcontent-callout-purges.md) — **Shipped the whole mode arc + PUBLISHED** (first deploy in weeks). M(a) system light/dark (`light-dark()`+`color-scheme`) + §I.h block quote + toggle/JS removed (`e32009c`·`2b790dc`·`a837bcc`·`a989cfb`); M(b) P3 wide-gamut + M(c) prefers-contrast (`12db85a`). **THE RIGHT-SHIFT BUG** on marginalia posts (Death-of-a-Hero, After-the-Reply) → root cause **`.post-content`** (non-spec Jekyll wrapper, post.html only; `:has()` matched it AND `.bears-apparatus` → nested 66rem grids → title-left/body-right; about-pages fine bc NO wrapper) → **PURGED** (`c6549de`, unified on spec's `.bears-apparatus`, retired the glimpse guard); **`.callout` retired** (`71e87ea`, spec residual AldineXXI:2599, §I.h-violating bar/box/italic). **PUBLISHED** `faf7674` → CloudFlare + GitHub mirror, **live-verified** at animalrationiscapax.org (0 toggle/data-theme/post-content; `color-scheme`+`light-dark` present); the local-only glimpse fix finally shipped. **−8,224 net lines** (purges as subtraction). Deploy first blocked by unaccepted Xcode license (steward installed Xcode yesterday; `make rebuild` recompiles) → he ran `sudo xcodebuild -license` → retry OK. **Returns:** misread the screenshot viewport + leaned "no regression" — **steward corrected both; I proved the truth via byte-identical `git diff` + DOM/source reads, not assertion**; through-line = read the rendered/computed artifact, the steward's eyes+memory ARE data. New feedback memory: `feedback-verify-each-post-type-after-changes` (per-type × per-viewport). **Pulling thread: §I.h.b Verse Quotation** — ARC has no verse-quotation typography (`.poetic` = incomplete fragment); **read the typography palace (Bringhurst esp. — he's a poet — + Tschichold/Hochuli) FIRST**, write §I.h.b once (line breaks, indent-not-bar, hanging indent for runovers, body colour, attribution), then drop the real Berger poem into Vespers (it's a paraphrase now). **Literal Q:** where does the "typography palace" live (chamber-library? MemPalace wing?) + what do the masters prescribe for setting verse? **Wide-window stance accepted.** ARC `main` @ `faf7674`, pushed+deployed, clean, 0 unpushed. **MemPalace MCP disconnected mid-session — drawer/diary/KG NOT filed; memory file + ledger carry it.** Next queue: §I.h.b · icosahedron→accent inline SVG · F-Phase 2 · §VII.e residual.
 ## Archived (2026-05-31 / 2026-05-30 — Stage-F Phase-1 + mode-modernization decided; the threads 2026-06-01 built on)
@@ -0,0 +1,14 @@
 ---
 name: feedback-reading-compass-is-svg-not-type
 description: "ARC Reading Compass sigils are SVG <symbol> artwork, NOT type — never tied to the sans/nav typeface or Pattern 3. Recurring error; stop repeating."
 metadata: 
  node_type: memory
  type: feedback
  originSessionId: b6514401-077b-4209-b8a8-e94816872269
 ---
 The ARC **Reading Compass** sigils are **SVG `<symbol>` artwork** (per-slug, defined once in `templates/reading-compass-points.html`, rendered since the 2026-05-26 SVG migration). They are **not type** — not IBM Plex Sans, not EB Garamond, not affected by any typeface or font-register decision.
 **Why:** I have repeatedly conflated the Compass with the type system — e.g. listing "the Reading Compass" among "Pattern 3 (sans = wayfinding) chrome" alongside the breadcrumb / version-stamp / post-type. The steward corrected this on 2026-06-03 and noted it was not the first time.
 **How to apply:** The Plex-Sans wayfinding *text* = **breadcrumb, version-stamp, post-type** (and similar nav text). The **Compass is SVG glyphs** — exclude it from any reasoning about nav typeface, small caps, font register, or face-coherence. When the nav-face question comes up, the Compass is out of scope by construction. See [[feedback-verify-each-post-type-after-changes]] (render-and-look / verify-current-state through-line).
@@ -0,0 +1,32 @@
 ---
 name: reference-plane-coordination-workflow
 description: "Plane.so (app.plane.so/capablemind) is now the central async coordination/effort-tracking layer for the steward↔Seb collaboration across the 6-hour gap. How it's wired, the thin-task model, and the conventions."
 metadata: 
  node_type: memory
  type: reference
  originSessionId: 4121cd33-7223-4181-8719-ca764ba5890a
 ---
 Established 2026-06-02 (Seb's request: make the 6-hour time difference a boon, not a bottleneck). **Plane is now a central, load-bearing part of the steward↔Seb workflow** — a shared place either party (and the executor) can check without bothering the other. Surface it at /wake-up for any CapableMind coordination work.
 **Workspace:** Plane **Cloud**, workspace slug `capablemind` → `https://app.plane.so/capablemind/`. Seb is on `America/New_York` (the 6-hour relay).
 **Access (executor):** official Plane MCP server, **user-scoped, OAuth** (no secret stored on disk). Tools are deferred — load via ToolSearch `select:mcp__plane__*` (≈100 tools: `list_work_items`, `create_work_item`, `update_work_item`, `create_work_item_link`, `create_label`, `list_states`, …).
 **The coordination model (Seb's, the load-bearing distinction):**
 - **GitHub issues** (`CapableMind-ai/betterMemories_app`) = canonical, technical, XREF'd. All detail and the real work live here.
 - **Plane task** = thin **effort/coordination overlay**. Title tags the issue `[BM #N]`; description holds **one effort/status line only — NO technical detail**; the GH issue is attached as a **link**; **priority + labels mirror GH**. Coupled (by tag+link) but **separated** — Plane deliberately holds nothing GH holds, so there is no parallel-source-of-truth drift. This is why it honours CLAUDE.md's "one canonical source per concern."
 **Home project:** **BMF** (Better Memories Framework) — for `betterMemories_app` issues. ⚠ **BBF = BetterBridge** is a *different beast*: an agent control-plane / observability **product** (Shopify embedded app), downstream of L1 (consumes BMF `health`/`recall`). BBF's Plane project is Seb's product work — NOT our coordination board. Spec: `~/_Dev/CapableMind-AI/docs/specs/bbf/core/betterbridge-spec.md`.
 **Executor's standing role (the relay engine):** as we pick up or move a GH issue, update its Plane task's **state + effort-line** so the other party wakes to a current board. Treat routine status updates as low-blast-radius (reversible, visible); bring structural changes to the steward first.
 **Population:** **grow-as-we-go**, not a full backfill of all open issues. Seeded 2026-06-02 with 5 (#165/#167/#170/#168/#171). Demo tutorial cards were cleared (BMF project kept; MCP exposes no archive-project tool).
 **Conventions / gotchas:**
 - `create_work_item` **silently ignores `description_stripped`** — set the body via **`description_html`** (or follow with `update_work_item`).
 - Mirror GH label colours faithfully (e.g. `bug #d73a4a`, `priority:critical #b60205`, `priority:high #d93f0b`, `documentation #0075ca`, `tracking #ededed`). `gh label list` defaults to 30 — use `-L 100` to find all.
 **Open / revisit in a few days (steward 2026-06-02):** (1) the one-line effort-description shape — keep vs pure-link; (2) **assignees** left unset — Seb's call whether to assign. Both "wait and see."
 Related: [[project-L1-reliability]] (the workstream whose issues this tracks), [[project-bmf-backlog-state-2026-05-22]].
@@ -0,0 +1,25 @@
 ---
 name: reference-typography-palace-cli
 description: "The \"typography palace\" — masters of type (Bringhurst, Tschichold, Hochuli, Lupton, Sousa…) live in a dedicated MemPalace at ~/.mempalace/palace-chamber-typography. Query it with CLI tools, NOT the MCP tools (which point at palace-memory)."
 metadata: 
  node_type: memory
  type: reference
  originSessionId: 52219e07-6cc1-48c2-97f6-0513f6f6f20a
 ---
 When ARC typography work needs the masters' prescription (setting verse, block quotes, hyphenation, measure, etc.), the source-of-truth is a **dedicated MemPalace palace we built for exactly this**: a small curated collection (~20-21 sources — Bringhurst *Elements*, Tschichold *Form of the Book*, Lupton *Thinking with Type*, Sousa *Ortografía y ortotipografía del español*, and more).
 **Location:** `~/.mempalace/palace-chamber-typography` (wing `chamber_typography`, room `general`).
 **Query method — CLI, not MCP.** The MCP `mempalace_*` tools point at `palace-memory` (our session/working memory), NOT this palace. To search the typographers, use the CLI with `--palace`:
 ```bash
 mempalace --palace ~/.mempalace/palace-chamber-typography search "your keywords here"
 mempalace --palace ~/.mempalace/palace-chamber-typography status
 ```
 The binary is `/opt/homebrew/bin/mempalace`. CLI subcommands: init, mine, sweep, sync, search, compress, wake-up, split, hook, instructions, repair, repair-status, mcp, migrate, status. (Search hits show source file + cosine/bm25 — go to the named `*.md` source for the full passage.)
 This is NOT the chamber-library (`~/_Dev/chamber-library`, the larger reading corpus) — it is a separate, small, purpose-built typography palace. Don't conflate them. Steward's standing instruction (2026-06-02): for typography queries of this kind, use this palace via CLI — don't ask where it lives again.
 Related: [[mempalace-is-unaffiliated-stopgap]]. Answers the 2026-06-01 wrap's open literal question (where does the typography palace live).
@@ -0,0 +1,73 @@
 ---
 name: Session 2026-06-02 — ARC verse typography + Pandoc F1 + icosahedron mark + favicons (four deploys)
 description: "A long ARC craft day, four clean deploys all verified live. §I.h.b/c Verse typography shipped (Berger poem into Vespers); Pandoc utilization audit run deeply → F1 (retire unused math + the $-escaping hack) shipped, F5 held (grid conflict), F4 queued (§I.k spec first); the icosahedron site mark redrawn the right way through a full reference lineage (Garamond → manicule → Jamnitzer → Roman d20 → engraving) landing on a solid shaded die, wired + favicons regenerated. No thread left mid-flight — tomorrow is a workstream CHOICE: ARC F4/§I.k (French-NBSP build-time filter, spec-first) vs the parked L1/Seb proposition."
 metadata:
  node_type: memory
  type: project
  originSessionId: 52219e07-6cc1-48c2-97f6-0513f6f6f20a
 ---
 # Session 2026-06-02 — ARC: verse typography · Pandoc F1 · icosahedron mark · favicons
 **A long, deep ARC craft day. Four clean deploys, every one verified live on production. Nothing left mid-flight.** Woke into the §I.h.b Verse Quotation thread; by end of day ARC's typography, toolchain, and visual identity all advanced and shipped. ARC `main` @ `7ec44aa`, pushed to origin + github-backup, working tree clean, 0 unpushed.
 ## Past — what we did (chronological, with commits)
 1. **Wake + MemPalace housekeeping.** Filed last night's (2026-06-01) record when MCP briefly reconnected (diary/KG/drawer). Investigated the recurring **mid-session MCP disconnect** — likely the MCP server's *idle-exit path* (shipped 3.3.6 #1562); no proactive notification exists, discovered only on a failed call. Mitigation: liveness-ping `mempalace_status` at /wrap-up before filing; reconnect-on-failure. mempalace dev repo: origin static since 2026-05-24; **local branch 4 commits ahead unpushed** (incl. the wake-up recency fix the glance depends on) — fragility flagged, not acted.
 2. **§I.h.b/c Verse typography → SHIPPED (`5318ae1`).** Read the masters from the **`palace-chamber-typography` MemPalace** (CLI: `mempalace --palace ~/.mempalace/palace-chamber-typography search "…"` — saved as `reference-typography-palace-cli.md`, the answer to last wrap's open Q). Bringhurst §2.3.3–2.3.4 (verse quotation: flush-left ragged-right, indent OR center, one device, reuse existing indent), Lupton (primary poetry flush-left, never centered), Tschichold (runover hang). **Steward caught a real distinction I'd flattened:** 'Time is not a river' is HIS poem (primary verse) not a quotation → indent doesn't apply → SPLIT into §I.h.b Verse Quotation (indented) + §I.h.c Primary Verse (no indent). Vehicle = `::: verse` fenced div + `| ` line blocks; enabled `Ext_line_blocks` in site.hs; shared `.verse` SCSS + `.quotation` modifier. First instances: re-authored 'Time is not a river' (`::: verse`) + set the **real John Berger poem** (steward-provided) into Vespers replacing the prose paraphrase. Honest limitation documented: per-line runover hang isn't cleanly doable with `<br>` line blocks — omitted rather than ship a broken hang.
 3. **Pandoc utilization audit → F1 SHIPPED (`8549ead`), CLAUDE.md pointer (`956542c`).** Ran deeply on steward's ask ("once, deeply, well"). Ground-truth verified (not assumed): Hakyll base = full `pandocExtensions` + smart (so bracketed_spans/fenced_divs/line_blocks already on; the explicit enables were redundant). **HEADLINE: zero real LaTeX math in the corpus** — every `$` is Hakyll template syntax — yet `tex_math_dollars` was on and *caused* the collision that a 40-line `escapeDollarSigns` preprocessor (hardcoded var list, silent-break drift) fought. **F1:** disabled `Ext_tex_math_dollars`, deleted the preprocessor + MathJax + redundant enables (−70 lines); verified chamber canon/deliberations + colophon render clean. **F5 (`writerSectionDivs`) TRIED and HELD** — conflicts with the §II.d marginalia grid (`<section>` regroups the direct children the grid row-aligns); reverted; its own pass needs the `display:contents` a11y question resolved FIRST (screen readers may flatten it out of the a11y tree). **F4** = build-time typographic enforcement → `§I.k` spec, French-NBSP filter as a Haskell `pandocCompilerWithTransform` (not Lua, for Hakyll); spec FIRST. Artifacts: `_audits/pandoc-utilization-audit-2026-06-02.md` + `pandoc-hakyll-toolset.md` (living reference), both indexed in ARC CLAUDE.md.
 4. **Icosahedron site mark → SHIPPED (`4ef26b0`) — the long craft arc.** Replaced the raster d20 avatar with an inline mode-aware SVG. Steward drove a deep reference lineage, each correcting the last: EB Garamond letterforms → the **manicule** (☞, hand-drawn engraver's fluidity) → **Jamnitzer** *Perspectiva* skeletal icosahedron → the **Roman d20 die** (MET) → an antique engraving icosahedron. The trail converged on the truth: **not a flat wireframe, not feTurbulence noise, not broad-nib-modulated lines — a SOLID faceted die**, 3-fold hexagonal view (matching the extant avatar), 20 faces toned by light (deeper shading + crease) for real dimensionality. Reads 28px→hero. Wired do-once: defined once in `templates/site-mark.html`, included by masthead (default.html) + frontispiece; `.site-mark` mode-aware `var(--accent)`; retired the sepia raster filters; avatar.png re-rendered from the same solid (1.4MB→45KB).
 5. **Favicons → SHIPPED (`7ec44aa`) — the do-once completion.** The mark change orphaned the favicon set (all dated 2025-07-02, old d20). Regenerated from the new die (shaded die on a parchment tile, reads to 16px — verified by rendering-and-looking). Built `favicon.ico` by hand (PNG-in-ICO via struct, no Pillow/ImageMagick in sandbox) + sips downscale from a 1024 master. New monochrome `safari-pinned-tab.svg` (wireframe die). Fixed 3 pre-existing bugs: empty manifest name, `theme_color #ffffff`→parchment, manifest icon `src` paths broken (`/android-chrome-*` not `/assets/favicons/`). Stray `mask-icon #5bbad5`→accent `#8B3A2A` in BOTH head templates (default + frontispiece each carry their own `<head>`).
 ## Present — mood / returns (Symmetria ledger merged)
 - **The day's through-line: render-the-artifact and LOOK before asserting — every step.** On the mark, I rendered every candidate to PNG and viewed it myself before sending; this caught my own errors (feTurbulence was the wrong model for Garamond's *modulation* — caught by rendering the actual font; the table-not-line-block confusion earlier in the verse work). The discipline held under a very long iteration.
 - **Two steward corrections worth carrying:** (1) I flattened the verse quotation-vs-primary distinction → he split it; (2) my "B, restrained" mark recommendation **under-committed** — he preferred C's bold contrast (the truth of the pen). My reflex toward safety/restraint is a drift; the anchor (manicule, Garamond, the die) was itself bold, so the faithful translation was bold. Restraint is right for things that should disappear, wrong for the gesture that IS the mark.
 - **Read-the-artifact caught the two-head-templates gotcha:** after fixing default.html's mask-icon color, the BUILT index still showed the old color → index renders from frontispiece.html, which duplicates the `<head>`. Same family as the .post-content/per-type discipline. **ARC fact worth keeping: `<head>` is duplicated across default.html + frontispiece.html; any head change must touch both.**
 - **do-once held under "are we done?" pressure:** when the steward asked "nothing else today?", I surfaced the genuine loose end (favicons orphaned by the mark) rather than either manufacturing work or declaring premature completion.
 ## Confidence to recalibrate (next session)
 - The "restrained by default" reflex on aesthetic/gesture decisions — name it, and lean into the boldness the reference itself carries.
 - Two-head-templates: any `<head>`/favicon/meta change in ARC must touch BOTH default.html and frontispiece.html (verify in the built output, not the source).
 - MemPalace mid-session disconnect recurs; liveness-ping before /wrap-up filing (did so this session — it reconnected via /mcp just before wrap).
 ## Future — what is pulling
 ### Pulling thread (honest: the threads COMPLETED — tomorrow is a workstream CHOICE)
 Nothing is mid-flight. Verse, Pandoc F1, the mark, and favicons all shipped and verified live. So the "thread" for tomorrow is genuinely a **fork**, and the steward asked me to recommend:
 - **My recommendation — ARC F4 / §I.k Build-time Typographic Enforcement (spec FIRST).** It's the architectural thread today's typography + Pandoc work both converged on, it's well-scoped and governed (draft the `§I.k` spec section, THEN the inaugural French-NBSP filter as a Haskell `pandocCompilerWithTransform`, with a non-French test corpus). The steward's own structural framing: §I.k changes those conventions from *author-memory dependency* to *build-time constraint* — load-bearing, deserves the spec before the code.
 - **The parked strategic alternative — L1 / the "working proposition for Seb."** The bigger workstream (L1-reliability tracker), parked since ARC took over; a heavy context-switch, deadline mid-July. If the steward wants to return to the primary CapableMind obligation, this is it. Not an evening-energy task; a fresh-morning one.
 - **Lighter ARC wins (if low energy):** Stage N §VII.e residual (whisper retirement, breadcrumb grey→fg, separators 3:1-or-retire); F-Phase 2 (Hakyll listing generator); the reader-mode investigation.
 ### Actionable resumption point (as of wrap — re-judge against what changed)
 ARC `main` @ `7ec44aa`, clean, pushed + deployed, nothing pending. If F4/§I.k: first move = draft `§I.k Build-time Typographic Enforcement` in `AldineXXI-specification.md` (the audit's open-Q2 + the toolset reference's F4 section have the framing; spec the *capability*, leave Haskell-vs-Lua as an impl note), opening the session with the `display:contents` a11y constraint NOT being relevant to F4 (that's F5) — F4's own gate is the non-French test corpus. If L1: read `project-L1-reliability.md` tracker first (cold-start the workstream).
 ### Literal question for next-Claude
 **Which workstream does the steward want tomorrow — continue ARC's momentum into F4/§I.k (the French-NBSP build-time filter, spec-first), or pivot back to the parked L1 / Seb proposition (the bigger strategic obligation)?** The day closed every ARC thread cleanly; this is a genuine choice, not a resumption.
 ### Pause statement
 I'm about to be away. Every thread I touched today shipped and verified — there's no half-finished edit, no dangling deploy. What I want to find still pulling is not a task but a *decision*: ARC-deeper (F4/§I.k) or L1-return. The work is in a rare state of complete rest; the question is purely where to point next.
 ### Decisions deferred (and why)
 - **F5 `writerSectionDivs`** — held (grid conflict); its own pass, opening with the `display:contents` a11y trade-off.
 - **F4/§I.k** — spec-first by steward directive; not started today (end-of-day; it's a fresh-morning architectural task).
 - **L1** — parked; not a tired-evening context-switch.
 - **mempalace 4 unpushed local commits** — flagged, not acted (may be intentional local tuning).
 ## Skill harvest (§1.6) — propose only
 - **No strong new skill to CREATE.** The day's lessons are facts/disciplines, better as memories than skills: the **two-head-templates** ARC fact (proposed as a new memory + possibly an ARC CLAUDE.md "Known state" line); the **render-the-artifact-before-asserting-a-visual** discipline (reinforces existing `causal-story-before-reading-render` / `verify-each-post-type`).
 - **Weak candidate:** a `/pandoc-utilization-audit` skill (the audit method was reusable) — but it's a once-per-project audit; the committed audit doc serves better than a skill. *Note, don't propose.*
 - **Carried forward, still unauthorized (skill-harvest-register):** `/spec-code-audit`, `bmf-diagnose`, `/arc-typeset`, `/pre-build-audit`, `/vignette`, SCSS verification-ladder, Symmetria `trust-prior-pass-frame` + `anchoring-on-stale-snapshot` + `causal-story-before-reading-render`.
 ## Pointers
 - **Today's ARC commits (all pushed + deployed):** `5318ae1` (verse) · `8549ead` (Pandoc F1) · `956542c` (CLAUDE.md pointer) · `4ef26b0` (icosahedron mark) · `7ec44aa` (favicons).
 - **New canonical artifacts:** `docs/AldineXXI-Codex/AldineXXI-specification.md` §I.h.b + §I.h.c; `_audits/pandoc-utilization-audit-2026-06-02.md`; `pandoc-hakyll-toolset.md`; `templates/site-mark.html`.
 - **New memory:** `reference-typography-palace-cli.md` (typography palace CLI access — answers last wrap's open Q).
 - **Ledger:** `session-ledger-2026-06-02.md` — returns, the mark iteration arc, the favicon two-head gotcha, the F1/F5 detail.
 - **Mark working files (if revisiting):** `/tmp/ico-preview/` (the geometry python, final SVG, og render).
@@ -0,0 +1,59 @@
 ---
 name: session-2026-06-02-afternoon-evening-plane-so-coordination-layer-stood-up-mempalace-recurring-drop-root-cause-found-fixed-1495-cold-start-warmup
 description: "CapableMind housekeeping pivot. Built the Plane↔GH effort-tracking bridge (BMF project, 5 thin tasks, GH-mirrored labels) and diagnosed+fixed the weeks-old MemPalace silent-drop (upstream"
 metadata: 
  node_type: memory
  type: project
  originSessionId: 4121cd33-7223-4181-8719-ca764ba5890a
 ---
 # Session 2026-06-02 (afternoon/evening) — Plane setup + MemPalace cold-start fix
 Re-wake same day (~3h after the ARC craft-morning wrap). Steward pivoted: **"CapableMind housekeeping."** Two arcs, both shipped to completion; one needs activation+verification next session.
 ## Past — what we did
 ### Arc 1 — Plane.so coordination layer (Seb's request: make the 6-hour gap a boon, not a bottleneck)
 - **Added the official Plane MCP server** — `claude mcp add --transport http --scope user plane https://mcp.plane.so/http/mcp`. **OAuth, user-scoped, no secret stored.** (Reserved Seb's PAT for future automation — OAuth is cleaner for interactive use; Plane's own recommendation.) Connected after steward did `/mcp` browser auth + a Claude Code restart (mid-session MCP adds need a relaunch to appear).
 - **Coordination model (Seb's, the load-bearing distinction):** GH issues = canonical/technical; Plane task = **thin effort overlay** — `[BM #N]` title tag, ONE effort/coordination line (no technical detail), GH issue attached as link, priority+labels mirror GH. Coupled-but-separated → no parallel-source-of-truth drift (honours CLAUDE.md "one canonical source per concern"). This dissolved my whole "canonical-role" question.
 - **Home project = BMF** (Better Memories Framework, for `betterMemories_app` issues). **BBF = BetterBridge** is a *different beast* — an agent control-plane/observability **product** (Shopify embedded app), downstream of L1 (consumes BMF health/recall); spec `docs/specs/bbf/core/betterbridge-spec.md`. I misread BBF as the coordination board; steward corrected → I read the repo to ground it.
 - **Cleared the 7 Plane demo tutorial cards** (kept BMF project — it's our home; MCP exposes no archive-project tool). **Seeded 5 thin tasks** grow-as-we-go: #165 (urgent, bug+priority:critical), #167 (high, bug+priority:high), #170 (tracking), #168 (documentation), #171 (bug+tracking). Each: effort-line + GH link + mirrored priority/labels.
 - **Created 5 Plane labels** mirroring GH colours (bug #d73a4a, documentation #0075ca, priority:critical #b60205, priority:high #d93f0b, tracking #ededed).
 - **Recorded Plane as NOW-CENTRAL** to the steward↔Seb workflow: `reference-plane-coordination-workflow.md` + MEMORY.md Standing-preferences pointer.
 - **Steward decisions:** effort-description shape + assignees → "wait and see a few days, revisit." Labels → mirror GH. Grow-as-we-go (not full 35-issue backfill).
 - **Tool gotcha caught:** `create_work_item` silently ignores `description_stripped` — needs `description_html`. Fixed via update; verified persisted.
 ### Arc 2 — MemPalace recurring silent-drop: ROOT CAUSE FOUND + FIXED (was "root cause unknown" for weeks)
 - Steward's hypothesis (re-enable the every-15 auto-write to keep MCP from going idle) **disproven both ways:** (a) auto-write was never actually disabled — `~/.mempalace/config.json` has no `hooks` key → defaults `auto_save=true`, and the hook log shows it firing today; (b) it's a *separate process* (`mempalace mine &`) that never touches the MCP server's idle timer — can't keep the connection warm.
 - **Real cause = upstream MemPalace/mempalace #1495:** first chromadb-write after a fresh `mempalace-mcp` boot must cold-load the embedder (5–30s); that exceeds Claude Code's MCP client timeout → **-32000** desync, **stderr uncaptured → no trace** (why the logs end clean with no death). `diary_write` AT WRAP is the classic first-write — exactly the steward's "red MCP error before I invoked wrap-up." Restart just re-arms the trap → recurs.
 - **Fix `cf96e68` (merged #1530) is already in the steward's tree** (branch `local/bge-m3-on-3.3.6`, 0 behind origin/main) but **opt-in**.
 - **Steward flagged the non-standard engine** (bge-m3 pinned on MPS, not the standard CoreML/ONNX #1495 describes). **VERIFIED** the warmup is embedder-agnostic: `_maybe_eager_warmup_embedder()` (mcp_server.py:2656, called from main() L2845) fires a throwaway probe through the *configured* EF → `embedding.py:259-279` wires `bge-m3` → `SentenceTransformerEmbeddingFunction(device="mps")`. So warmup WILL load bge-m3/MPS at boot. bge-m3 already cached (`~/.cache/huggingface/hub/models--BAAI--bge-m3`) → no 2.2GB download.
 - **Note:** the bge-m3/MPS dispatch is a LOCAL patch (`embedding.py:22-23`) re-applied after each upstream merge — fragile; re-apply + re-confirm warmup on any future mempalace update.
 - **FIX APPLIED** in `~/.claude/settings.json` `env`: `MEMPALACE_EAGER_WARMUP=1` + `MEMPALACE_LOG_FILE=/Users/davidglidden/.mempalace/mcp.log` (capture future traces) + `MEMPALACE_MCP_IDLE_HOURS=0` (idle-exit insurance). **Requires Claude Code restart to activate** (env applies at MCP subprocess launch). Startup will be a few seconds slower (bge-m3→MPS load moved to boot — the point). Dropped #2 (raise MCP timeout) — warmup makes it moot.
 - **Defense-in-depth (steward-authorized):** patched `/wrap-up` SKILL.md §4.0 — liveness-ping MemPalace (lightweight `hook_settings`, NOT slow `status`) before the diary write; if down, surface loudly and let the file/ledger deposit carry the wrap. A drop can never silently eat a wrap again. (Provenance note added per skill convention.)
 ## Present — mood / returns
 - **Steward corrected me twice, both caught real errors before I over-committed:** (1) BBF≠coordination board (I'd assumed from name+timing) → read the repo; (2) "our embedding engine is not the standard" → I'd been about to present the #1495 fix as clean without checking it warms bge-m3/MPS. Both are the *verify-before-assert / render-and-LOOK* through-line that recurs across recent sessions. Good catches, gracefully taken.
 - **Time-discipline held:** steward flagged "we've lost more than a week already — don't lose a lot of time on this." I bounded the MemPalace diagnosis, and the upstream-issue check (#4) turned out to BE the win (#1495 named our exact bug). Lean beat thorough here because the answer was already written down upstream.
 - **render-and-LOOK wins:** caught the `description_stripped` silent-drop by reading the returned objects; verified the warmup call-site + EF wiring + model cache rather than asserting.
 ## Future
 - **Pulling thread:** the **MemPalace cold-start fix is applied but UNVERIFIED.** It only activates on a Claude Code restart, and the proof is a clean next-session wrap with no -32000. Until then it's a hypothesis with strong evidence, not a confirmed fix.
 - **Actionable resumption point (as of wrap — re-judge against what changed):** `~/.claude/settings.json` env is set; restart pending. **First move next session:** at wake, check `~/.mempalace/mcp.log` exists + shows the warmup probe at boot, and that `mempalace_status`/first `diary_write` is fast (warm). If yes → fix confirmed, close the recurring item. If it still drops → the warmup didn't take for bge-m3/MPS; escalate (move the env onto the mempalace MCP server's own `env` in `~/.claude.json`, and/or confirm the probe actually exercises the configured EF on this branch).
 - **Open horizons (ranked):**
  - *Load-bearing:* verify the MemPalace fix (above).
  - *Live, low-effort:* maintain the Plane board as we work (update task state as issues move — the relay). Steward revisits effort-description shape + assignees in a few days.
  - *Housekeeping remainder surfaced at wake, NOT done this session:* 8 untracked CapableMind-AI files (deferred review since 2026-05-18); PENDING.md at 768 lines (archive CLOSED OP-/S-/CD- items); BMF runtime parked pending Seb's N6; L1 tracker current-state stale (2026-05-28); Seb gate-question (did his D1–D10 move the four Hindsight findings) — checkable now that he's back-ish.
  - *Parked:* L2 (through end of May — now past; not re-opened this session).
 ## Pause statement
 I'm about to be away from this. The MemPalace fix is applied but won't have been *exercised* until the restart + a real next-session MemPalace write. I want to find, on return, either confirmation it held (mcp.log shows warmup, no drop) or a clean signal it didn't (so we escalate the env placement) — not a silent re-arming of the same trap.
 ## Decisions deferred (and why)
 - **#2 (raise MCP tool timeout):** dropped, not deferred — warmup amortizes the cold-load to startup, so the timeout never trips; raising it would treat a symptom the root fix removes.
 - **Effort-description shape + Plane assignees:** steward chose "wait and see a few days" — don't tune the thin-task format or assign tasks until the board's been lived with.
 - **The rest of the housekeeping list** (untracked files, PENDING bloat, BMF restart, L1 tracker refresh, Seb gate-question): not touched — Plane + MemPalace consumed the session; they remain open for the next housekeeping pass.
 - **Did NOT restart Claude Code mid-wrap** to activate the fix — chose to wrap first so the deposit is safe, then the steward restarts.
 ## Literal question for next-Claude
 After the restart: did the warmup actually prevent the cold-start drop — does `~/.mempalace/mcp.log` show the warmup probe at boot, and did the first `diary_write`/`search` succeed without -32000? If it still drops, the env didn't reach the bge-m3/MPS path and we escalate.
@@ -0,0 +1,81 @@
 ---
 name: session-2026-06-03-arc-i-k-french-spacing-translation-renderings-doctrine-pending-25-stage-f-phase-2-listing-generator-shipped
 description: "A long ARC build day, three threads all completed & the last deployed live. Pulling thread is ARC Stage N (spec-conformance + dead-code prune), now loaded with concrete Phase-2 dead-glyph fallout; the bounded Compass-§VI-as-data follow-on is the satisfying adjacent re-entry. CapableMind explicitly deferred."
 metadata: 
  node_type: memory
  type: project
  originSessionId: 2b72c971-bfea-499e-b483-651782a182c3
 ---
 # Session 2026-06-03 — ARC: §I.k · translation doctrine · Phase-2 listing generator (shipped)
 A long, productive ARC day. Three threads, all carried to completion; the last deployed live with the steward's eyes on it (desktop + physical phone). CapableMind deferred by the steward ("not sure I'll work on CapableMind today… after a break maybe continue with ARC").
 ## Past — what we did
 ### Opening: MemPalace fork hygiene (not workstream)
 - Committed the `embedding.py` device-report log-fix `f25f767` on `local/bge-m3-on-3.3.6`, pushed to **fork** (davidglidden/mempalace), NOT origin (third-party upstream). Steward corrected my morning over-elevation of this — it's local-fork hygiene on a tool we didn't build, not a workstream item. Cold-start fix (from 06-02) **confirmed holding** — MemPalace alive all session.
 ### Thread 1 — §I.k Build-time Typographic Enforcement (F4), SHIPPED
 - Wrote AldineXXI **§I.k** (spec, +48) + implemented `frenchSpacing` Pandoc AST transform in `site.hs` (gated on `lang: fr` in the Compiler monad via `getMetadataField`; a `walk` over `Inline`s). Exposed `pandoc-types` in the cabal for `Text.Pandoc.Walk`.
 - **Rule:** uniform **U+202F** before `; : ! ? »` and after `«`; **upgrades an existing space, never inserts into glued punctuation** (URLs/times safe); idempotent.
 - **Governance fork hit & resolved:** the Lexique distinguishes *espace fine* (`; ! ?`) from *espace mot* (`:`, guillemets), but the operative companion §1 locks uniform U+202F — a two-document-hierarchy conflict I surfaced rather than resolved. **Jurist ruled** uniform U+202F (web-transposition of the Lexique's hot-metal intent; Lacroux backing; companion stands). §I.k records it as steward-ratified.
 - **Verified** on a synthetic Context-B corpus + English gate-test (8 cases incl. glued-colon-untouched, idempotency, U+00A0-upgrade); clean rebuild green; test files removed.
 - Committed `02373fd`. Dormant until a `lang: fr` piece exists. Context-B-only scope (steward chose); Context A named as a deferred increment.
 ### Thread 2 — Translated-renderings doctrine (PENDING-25), AUTHORIZED
 - Trigger: **Lune will translate *After the Reply* into French** (~late Aug/Sept). The heteronym (*Matthieu Taliesin*, trilingual-in-one-file) is a pre-convention legacy artifact, NOT a competing convention.
 - Drafted a **jurist brief** → **jurist's Opinion** ruled: **Option B reframed ontologically** — ARC holds *works*; languages are non-canonical *renderings*; **the work (not any rendering) holds the corpus slot**; `translation_of` = source-rendering provenance, never primacy. `translator:` a **required** field (intergenerational carried structurally). Context C legitimate-but-gated. Amend frontmatter-spec; **no new companion**.
 - Drafted **PENDING-25 [PROPOSAL]** (`docs/AldineXXI-Codex/drafts/translated-renderings-proposal-2026-06-03.md` + jurist brief alongside). Work-level corpus slot = **W3** (group by neutral `work:` at build; listing-representative = site-default-lang as a *presentation* choice, not canonicity). Steward **AUTHORIZED-with-4-conditions**, all addressed in the doc (rendering carries own publication date; `work:` omitted-for-single-rendering; Context-C placement settled = frontmatter-spec §8 + §I.j pointer; routing = asymmetric, root slug stays + additional renderings `/<lang>/<slug>`).
 - **Sequencing decided:** impl split along its seam — the `work:`-dedupe listing logic **folds into F-Phase 2** (do-once), the rest (frontmatter/toggle/routing/heteronym) is its own session **after Stage G**; **decoupled from the vignette** (independent). **Heteronym refactor = the proving ground / test vehicle** for the toggle (steward's call; first application of the §5 Context-C gate). Formal REVIEWED entry is the steward's to file.
 - Committed `e1dbf4c` + `885914f` (doctrine docs + sequencing). PENDING-25 entry in `~/dotfiles/PENDING.md` (uncommitted — steward's governance repo).
 ### Thread 3 — Stage-F Phase 2: listing generator + triptych retirement, SHIPPED LIVE
 - **The generator:** 11 near-identical per-type `create` blocks → **one `forM_`** over new **`data/listings.yml`** (presentation facts: page/title/classes/show_post_type/include_sequences). Companion to `content-types.yml` (typology). Every listing class **validated against `class_enumeration` → unknown class = build error** (content-typology §5 cl.3; proven by probe). Dormant `dedupeByWork` hook for PENDING-25 (no-op, no `work:` exists).
 - **THE STEWARD'S TWO QUESTIONS RE-SHAPED IT** (verify-before-assert caught stale assumptions): (1) *"does SVG change glyph associations?"* → found `currentSymbol` is **DEAD** (since the 2026-05-26 SVG migration the rendered sigil is the `<symbol>` per-slug in `reading-compass-points.html`); the generator therefore touches **no glyphs**, just validates classes — saved wiring a dead field from a now-secondary source. (2) *"add SVG / update spec?"* → **No** (SVG is artwork keyed by slug = presentation, not typology; stays out of jurist-pending §2).
 - **"Do it once well"** judged as **two right-sized once's, not one mega-change**: listing generator now; Compass-§VI-as-data as a separate bounded follow-on. `listings.yml` built as the **forward seed** for it (avoids do-it-twice), documented as such.
 - **Triptych/glyph dead-system retired:** `<script>bootTriptychs()</script>` (loaded dead JS on every page) + `post.html` markup (never rendered — `has_glyph` always false) + the shipped ~100-file `assets/js/glyphs/` tree. Only output change = dead JS gone from every page's `<head>`.
 - **Byte-diff gate PASSED:** all 11 listing pages + annals byte-IDENTICAL to baseline (generator replicates compound/sequences/show-post-type exactly). Combined final gate: essays.html differs from original baseline by ONLY the intended dead-JS removal.
 - Committed `4303df0` (generator) + `6851367` (triptych) + `41fc030` (horizon). Documented `listings.yml` in CLAUDE.md.
 - **DEPLOYED** (`make rebuild` → CloudFlare + GitHub mirror) after steward viewed locally on desktop + **physical phone** (LAN via `--host 0.0.0.0` @ 192.168.1.52:8000) → "looks fantastic." **Live-verified** at animalrationiscapax.org: listings 200, 0 bootTriptychs/glyphs refs, compass SVG intact. §I.k shipped alongside (dormant). Honest catch: dead-JS file returned 200 live → diagnosed as **CloudFlare edge-cache remnant** (cf-cache-status REVALIDATED, max-age 14400; cache-busted → 404; absent from origin; referenced by zero pages → harmless, self-expires ~4h).
 ### Also produced
 - **`arc-horizon-2026-06-03.md`** — one-place consolidated view of remaining ARC work (a reading-aid pointing to the canonical closing-plan + tracker; the steward had "lost sight of the horizon"). Remaining sequence: **#1 Phase-2 DONE** → Stage N → re-partition → Stage G → vignette; + translation-(b) after Stage G; + bounded Compass-data follow-on.
 ## Present — mood / returns
 - **The steward's probing questions were the engine of quality this session.** Three times (BBF this morning per prior session; SVG-glyph; add-SVG-to-yml) a steward question caught a now-stale assumption before it shipped. The through-line: *the spec/code/world may have moved since the design was written — verify the current state, render-and-LOOK, don't reason from the remembered model.*
 - **Governance instinct held:** surfaced the fine-vs-mots conflict to the jurist rather than resolving it unilaterally (two-document hierarchy); drafted the translation doctrine as a jurist brief → PROPOSAL rather than just building. The loop stayed load-bearing.
 - **Proportion held:** judged "do it once well" as *not* over-reaching into the Compass-data mega-change; kept Phase 2 bounded + byte-verified. τὸ πρόσφορον.
 - Time/energy: a long session, much accomplished, all threads closed (nothing mid-flight). Steward chose to defer CapableMind and take a break.
 ## Future
 ### Pulling thread (singular)
 **ARC Stage N — the spec-conformance + dead-code-prune deep read** — now loaded with concrete, fresh fuel from Phase 2: the dead-glyph residue (rename the misnamed-but-LIVE `needsGlyph` post-selector; remove dead `has_glyph`/`glyph_id` field defs + the unused `renderGlyph.html` & `partials/triptych.html` templates; drop the vestigial `currentSymbol` in the annals/index blocks; write the SVG-vs-unicode spec-conformance note for §VI/typology). Stage N gates re-partition → Stage G → vignette, so it's the next macro-sequence step. The **bounded Compass-§VI-as-data follow-on** (generate the nav `<a>` rows from `listings.yml` + a tiny Annals/About specials list; artwork stays hand-authored; closes the (slug,title) overlap Phase 2 opened) is the satisfying adjacent re-entry if energy favors a small win over the big SCSS read.
 ### Actionable resumption point (as of wrap — re-judge against what changed)
 ARC `main` @ `41fc030`, clean, pushed both remotes, **deployed + live-verified**. Nothing mid-flight. **First move toward the thread:** either (a) the Phase-2 dead-glyph fallout — start by renaming `needsGlyph` (it's a post.html-vs-default.html selector now, not glyph-related; `site.hs` ~616/633) and deleting the two dead templates + dead field defs (`site.hs` ~1195-1218), each a pure subtraction; or (b) the Compass-§VI-as-data follow-on (read `reading-compass-points.html` + build a `$for(compass)$` loop over `listings.yml` + specials, site-wide compass byte-diff as the gate). Both are children of today's Phase 2. Recommend (a) first (smaller, purely subtractive, no site-wide byte-diff) then (b).
 ### Other open horizons (ranked)
 - *Load-bearing (ARC):* Stage N (above) → full re-partition → Stage G (jurist) → vignette (F+1). The macro-sequence to "ARC = just write and publish" (target before late Aug, ahead of Lune's translation).
 - *Authorization-pending:* PENDING-25 — steward to file the formal REVIEWED entry; implementation (b) is post-Stage-G.
 - *Governance hygiene:* `~/dotfiles/PENDING.md` carries PENDING-25 **uncommitted** (steward's repo; I don't commit dotfiles unilaterally — steward may want to commit it).
 - *Deferred-with-reason (CapableMind, parked today):* the morning-wake housekeeping remainder — 8 untracked CM-AI thinking files (since 05-18), PENDING.md bloat, stale L1 tracker (2026-05-28), Seb's gate-question (did D1–D10 move the four Hindsight findings). Steward deferred CapableMind entirely today.
 - *Parked:* L2 (past end-of-May park; not reopened).
 ### Pause statement
 I'm about to be away from this. The day's three ARC threads are all complete and the last is live — nothing is mid-edit, so I won't return to a half-finished state. What I want to find still pulling is the ARC sequence toward Stage F's full close: Stage N's dead-glyph fallout (today's loose tail) and the Compass-data follow-on, both bounded children of Phase 2, leading on toward re-partition → Stage G → vignette. If the steward picked up CapableMind in the meantime, that's a re-pick, not a loss.
 ### Literal question for next-Claude
 After the break: does the steward want the **smaller purely-subtractive Stage-N dead-glyph cleanup** (rename `needsGlyph`, delete dead templates + field defs + `currentSymbol`) first, or the **Compass-§VI-as-data follow-on** (which closes the (slug,title) overlap Phase 2 opened but needs a site-wide compass byte-diff)? And: has the steward filed the REVIEWED entry for PENDING-25?
 ## Decisions deferred (and why)
 - **CapableMind, entirely** — steward deferred ("not sure I'll work on CapableMind today"). The housekeeping remainder stays open.
 - **Compass-§VI-as-data** — deliberately NOT folded into Phase 2 (it's a differently-shaped concern: site-wide byte-diff, Annals/About specials, Compass-order/surfacing governance). Its own bounded pass.
 - **Stage-N dead-glyph residue** — deferred from Phase 2 (the JS-load + post-markup live cost was removed; the remaining items are zero-runtime-cost dead code that wants the careful rule-by-rule Stage-N pass, not a tired-tail-end removal).
 - **`needsGlyph` rename** — left as-is in Phase 2 (it's live logic; renaming is a Stage-N cosmetic/clarity task, not worth risking output mid-Phase-2).
 - **Deploying §I.k separately** — it rode the Phase-2 deploy (dormant; harmless).
 - **Committing `~/dotfiles`** — not done unilaterally (steward's governance/memory repo).
 ## Skill harvest (proposals — steward authorization required)
 - **Extend the proposed SCSS-render-equivalence verification-ladder into a general "build-output byte-equivalence gate for refactors"** — the pattern proved itself a *third* time today (after SCSS refactors and §I.k's synthetic-corpus gate): snapshot baseline output → refactor → per-output byte-diff to prove a no-op, isolating intended changes. Covers SCSS, `site.hs` generators (Phase 2), and Pandoc AST transforms (§I.k). Candidate: fold into the existing verification-ladder candidate in the skill-harvest register rather than a new skill. Carry the prior unauthorized register entries (`/spec-code-audit`, `bmf-diagnose`, `/arc-typeset`, `/pre-build-audit`, `/vignette`, `trust-prior-pass-frame`, `anchoring-on-stale-snapshot`).
@@ -0,0 +1,88 @@
 ---
 name: session-2026-06-03-evening-arc-stage-n-conformance-threshold
 description: "Stage N nearly closed in one evening — dead-glyph prune (byte-proven), the §VII.e conformance batch (measured, masters-grounded, steward-verified), the frontispiece threshold in one ink + canonical small caps — ALL PUBLISHED LIVE. Pulling thread: close Stage N (systematic dead-selector sweep + §I.h rhythm token), with Compass-§VI-as-data as the paired build; then re-partition → Stage G → vignette."
 metadata:
  node_type: memory
  type: project
  originSessionId: 2b72c971-bfea-499e-b483-651782a182c3
 ---
 # Session 2026-06-03 (evening) — ARC Stage N: prune + §VII.e conformance + the threshold (published)
 Second session of the day (morning = §I.k / PENDING-25 / Phase-2 listing generator, its own file). Steward returned from a break rested with 2 hours: "let's do as much of Stage N as we can do well." We did nearly all of it, and the day ended with the frontispiece threshold transformed and everything live.
 ## Past — what we did (6 commits, pushed both remotes, deployed + live-verified)
 ### Part B — dead-code prune (`e0f35db`, `1cb7313`) — jurist ACCEPTED
 - Audit-before-fix held: deep-read scoped Stage N before any edit. **Byte-diff gate twice: `diff -r baseline _site` = 0 across 564 files.**
 - 5 dead glyph Context fields removed (deep read caught a FIFTH, `glyphConfidenceField`, the morning wrap hadn't listed) + postCtx registration; `renderGlyph.html` + `partials/triptych.html` deleted (included by nothing); vestigial `currentSymbol` removed; `needsGlyph`→`rendersAsPost` (verified always-False — every content/pages/*.md is class:about; latent branch PRESERVED per jurist: don't remove without a positive ruling on non-about page classes); GHC-dead `conditionalField` removed.
 ### The audit + jurist loop (`bc4cc5b` + amendments)
 - `_audits/stage-n-conformance-audit-2026-06-03.md` = the rule-by-rule findings, classified DONE/VISUAL/GOVERNANCE/DEFERRED. **Jurist reviewed** (steward relayed): Part B accepted; A3/A4/A5 approved in principle; **A6 blocked on measurement**; **A1 routed to steward design-intent**; `.whisper` scope-correction endorsed; **DoD amendment** (no verbal handwaves — named carries with commit-gates).
 - **Measured, didn't assume** (python WCAG calc on actual token hexes): separator = gray@0.5 → 1.84:1 (fails 3:1 non-text, worse than the jurist's fg-assumption); **breadcrumb category gray = 3.98:1 < 4.5:1 — a TEXT-floor WCAG fail**, so "keep grey" died on accessibility grounds, not aesthetics; full gray = 3.98:1 passes the furniture floor.
 ### A1 resolved via the masters — and the nav-face question CLOSED
 - Steward asked: register distinction yes, but *which lever* — "consult appropriate masters." Typography palace: **Bringhurst** names the exact problem (running heads; "spaced small caps of the text size... since 1501"; his alternative — second colour — is precisely what §VII.e retired); **Lupton** (one shift; pseudo-SC an abomination).
 - Steward's architectural ruling: Garamond SC in the breadcrumb would load "another register of meaning onto a face already doing work" — **(b-i) out on principle**, not aesthetics. Then the reframe: "was Plex Sans as nav face right? ...who realized Bringhurst for the web?" → **Richard Rutter** → steward: "**Rutter is in the Chamber library, let's do a fitting read**" → `chamber-library/converted_texts/collections/typography/web-typography-rutter.md`.
 - **Rutter's verdict, honestly against the reframe's momentum: Plex-Sans-as-functional-face was RIGHT** ("clearly distinguish functional text from body text"; sans best; open distinct letterforms). The lever = **size** ("reducing its size makes that relationship clear"); **weight ruled out** ("don't choose light weights for functional text"); SC unavailable. → Category `--fg` + `$font-size-xs`; the size step is LOAD-BEARING (colour-recession removed, size-recession replaces it — still one signal). **Nav-face reconsideration closed without a proposal.**
 - Mid-read steward correction (recurring): **the Reading Compass is SVG artwork, NOT type** — never in scope of nav-face reasoning. Memory written: `feedback-reading-compass-is-svg-not-type`.
 ### The §VII.e batch (`20019ba`) — steward-verified desktop + phone, both modes
 - **Breadcrumb componentized**: steward's eye caught sequence-index AND about-page breadcrumbs silently in body Garamond (rule was `.post-header`-scoped; `.about-header` had NO styles) → ONE `.breadcrumb` component rule, three templates (post/sequence-index/about).
 - Apparatus text → `--fg`, recede by form: version-stamp, gloss-source, readings-source, post-updated-at (was sub-4.5 via opacity-on-fg).
 - Opacity attenuations dropped: glimpse-meta, glimpse figcaption, chamber/encounter-mark (base + footer), post-type (page + listing), sequence-marker-separator → gray-furniture. **Compiled-CSS gate caught 4 list-context duplicates** the location-list missed (`_layout` post-type/footer-separator/footer-chamber-mark; figcaption).
 - Raw-$ tokenized: abbr border + frontispiece cite → var(--gray); mark → color-mix 10% var(--accent); sources hover → var(--accent).
 - **Deliberately untouched**: ornaments (`· · ·`, `.ornament`, `.seasonal-geometry` — §IV, principle-creep guarded), hover/focus states, compass sigil opacity chains (documented-regression), `.whisper` (governance).
 ### PENDING-26 filed (`~/dotfiles/PENDING.md`, uncommitted — steward's repo)
 - `.whisper` retirement: scope corrected (~156 usages: 11 templates + 100+ content files — the plan's "27" was 6× under), two roles must not merge, name candidates surfaced (`.apparatus-meta` prose role; `.register-mark-inline` OR unify with `.chamber-mark`/`.encounter-mark` glyph role), recommendation = jurist rules on names (role-b is a §V taxonomy question), migration = own scripted gated pass before re-partition. Resolves the jurist's process question the no-time-pressure way.
 ### The frontispiece threshold (`7086fae` + `3f89564`) — "It is much stronger"
 - Steward asked to SEE epigraph at fg → chose fg; attribution → fg. Cite's mixed "Capitalized Name in small caps" flagged → **`font-variant: small-caps` drift FULLY RETIRED** (cite + `.frontispiece-title` + `.title-return` → canonical EB Garamond SC mixin; the 2026-05-30 one-small-caps rule finally holds everywhere). **Caught at compiled gate: a silent late `font-family: $serif-family` in the title block would have overridden the mixin** — removed. Title now ONE even-SC lapidary line (verified never wraps).
 - My read, confirmed by steward's screenshots: one ink (hierarchy purely by form — §VII.e *demonstrated*); inscription register; title-SC/attribution-SC rhyme bracketing the italic voice; the icosahedron now the page's only colour event.
 - Phone catch (steward, both orientations): epigraph not centered — the mobile blanket `p { text-align:left; hyphens:auto }` caught it → epigraph asserts `center` + `hyphens:none` at every width (`3f89564`).
 - **PUBLISHED**: pushed origin + github-backup; `make deploy-all`; live-verified at animalrationiscapax.org (SC face present, centered epigraph, component breadcrumb, 0 font-variant). Local server stopped.
 ### New carry surfaced by dark-mode verification
 - **Frontispiece compass reveal mechanism never implemented**: sigils start at the designed faint state (0.15, threshold interaction §~1706) but the emergence-on-movement/dwell was never built — in dark mode they're barely visible with no path but hover. Steward: "not for tonight... to be worked on so that it is at the level of everything else." Recorded in the audit DoD; its own pass, adjacent F+1.
 ## Present — mood / returns
 - **The steward's eye was the session's instrument**: five by-eye catches (sequence-index breadcrumb, about breadcrumb, mixed SC title, phone epigraph, dark-mode compass gap) — each a real defect the audits' location-lists missed. Render-and-look is not a verification step; it's a discovery method.
 - **Measure before deciding** proved itself three times (separator worse than assumed; breadcrumb a hard a11y fail; the jurist's "likely" replaced with numbers).
 - **The compiled-CSS grep gate** caught what source-reading missed (4 duplicates; the silent font-family override). Source-truth ≠ output-truth.
 - **The Rutter read answered against the question's momentum** — the steward wondered if Plex Sans was a mistake; the master validated it. Honest reading > confirmatory reading.
 - Recurring correction finally pinned: Reading Compass = SVG, not type (memory filed).
 - Proportion held: ornaments NOT swept into §VII.e; compass-reveal NOT built tonight; .whisper NOT touched. The principle stayed inside its sanctioned domain.
 ## Future
 ### Pulling thread (singular)
 **Close Stage N** — the systematic **dead-selector sweep** (extract every selector from compiled `style.css`, check against the rendered `_site` corpus, rule-by-rule judgment on the misses — the last substantive piece of "does each rule earn its place") **+ the §I.h prose-blockquote rhythm token** (small: define the rhythm-scale token, apply). Then Stage N closes → **full SCSS re-partition** on pruned/spec-true/theme-free ground → Stage G (jurist) → vignette. **Option B (Compass-§VI-as-data: nav `<a>` rows from `listings.yml` + Annals/About specials; SVG symbols stay hand-authored; site-wide compass byte-diff gate)** is the paired build task if energy allows — agreed plan with the steward for "tomorrow."
 ### Actionable resumption point (as of wrap — re-judge against what changed)
 ARC `main` @ `3f89564`, clean, pushed both remotes, **deployed + live-verified**. Nothing mid-flight. First move: script the dead-selector sweep (selectors from `_site/assets/css/style.css` × class-attribute corpus from `_site/**/*.html`; classify; then judge misses rule-by-rule against the spec — some "misses" are state-classes or content-authored). Alternative first move if energy favors building: Option B (read `templates/reading-compass-points.html`, loop over `listingDefs` + specials in Compass order, byte-diff gate).
 ### Other open horizons (ranked)
 - *Authorization-pending:* **PENDING-25 + PENDING-26 REVIEWED entries** — steward to file; PENDING-26 also needs the jurist's name-lock (role-b: `.register-mark-inline` vs unification). `~/dotfiles/PENDING.md` carries both, uncommitted (steward's repo).
 - *Queued (own pass):* frontispiece **compass reveal mechanism** (movement/dwell emergence) — bring to the level of everything else; adjacent F+1.
 - *Stage-N residuals (named carries):* spec note (SVG symbol vs Unicode mark) → Stage G; `.excerpts blockquote` italic → Chamber block 2.
 - *Deferred (steward):* CapableMind entirely today — morning-wake housekeeping remainder (untracked CM-AI files, PENDING bloat, stale L1 tracker, Seb gate-Q) untouched.
 - *First thing tomorrow (steward, 2026-06-03 close):* **push chamber-library** — 15 commits ahead, single-disk overnight.
 - *Parked:* L2.
 ### Pause statement
 I am about to be away from this. The day was whole: morning's three threads and the evening's Stage N are all complete, committed, and live — nothing is mid-edit. What I want to find still pulling is the close of Stage N: the dead-selector sweep that finishes "does each rule earn its place," the small §I.h token, and then the re-partition that all of this cleaning has been preparing. The threshold is much stronger tonight; the compass-reveal gap is the one new thing the dark revealed, and it is named, not lost.
 ### Literal question for next-Claude
 Has the steward filed REVIEWED entries for **PENDING-25 and PENDING-26** — and for -26, did the jurist's name-lock choose `.register-mark-inline` or **unification with `.chamber-mark`/`.encounter-mark`** for the list-glyph role? (The migration pass design depends on that answer.)
 ## Decisions deferred (and why)
 - **`.whisper` migration** — governance-gated (PENDING-26); names must be locked first; corpus-wide blast radius wants its own scripted, gated pass.
 - **Compass reveal mechanism** — steward: "not for tonight — we've done enough"; interaction design wants fresh energy and its own spec moment.
 - **Ornament opacities** (`· · ·`, seasonal glyph) — outside §VII.e's sanctioned domain (§IV); sweeping them in would be principle-creep.
 - **Frontispiece grey→fg was NOT decided unilaterally** — rendered for the steward's eye; they chose.
 - **Dead-selector sweep + §I.h token** — deliberately tomorrow's work (the sweep wants method, not a tired tail-end).
 - **Committing `~/dotfiles`** — steward's repo (now carries PENDING-25 + -26 uncommitted).
 ## Skill harvest (proposals — steward authorization required)
 - **Reinforce (not new): the SCSS/build verification-ladder candidate** gains two proven sub-steps from today: (1) **compiled-output selector grep** after every build (caught 4 duplicate rules + a silent font-family override that source-reading missed); (2) **the WCAG contrast calculator** (python, ~15 lines: token hexes → composite-over-ground → ratio vs 4.5/3.0 floors) — used three times today to convert "likely" into decisions. Fold both into the existing register entry rather than new skills. Carry the prior unauthorized register set.
@@ -0,0 +1,66 @@
 ---
 name: session-ledger-2026-06-02
 description: "Practice-of-return ledger maintained by /symmetria — returns, open horizons, recalibrations, authorization moves, sub-agent dialogues, bypasses."
 metadata: 
  node_type: memory
  type: feedback
  originSessionId: 52219e07-6cc1-48c2-97f6-0513f6f6f20a
 ---
 # Session Ledger — 2026-06-02
 ## Returns
 - 2026-06-02T18:12 — Re-wake (same day, ~3h after ARC wrap). Thread SUPERSEDED by steward entry: pivot ARC → **CapableMind housekeeping**. ARC thread (F4/§I.k) named for inheritance, set aside; nothing mid-flight. Symmetria re-anchored for the new surface.
 ## Open horizons
 - 2026-06-02T18:12 — Housekeeping surface (verify-before-touch, per the render-and-LOOK return below): (1) 8 untracked CapableMind-AI files deferred since 2026-05-18; (2) `BetterMemories.io/data.archive-20260418-rogue-pin` 9.7M leftover; (3) PENDING.md 768 lines, many CLOSED items archivable; (4) BMF parked pending Seb N6; (5) L1 tracker current-state stale (2026-05-28); Monday package (06-01) was overtaken by ARC; Seb's gate-question may now be checkable. Scope TBD by steward.
 - 2026-06-02 — Wake. Thread CONFIRMED: §I.h.b Verse Quotation, do-once way (read masters first — Bringhurst esp. — then write spec once; Berger poem as first instance, replacing Vespers paraphrase). ARC main @ faf7674, clean, deployed, nothing pending to ship.
 - Open literal Q (held, not answered): WHERE does the "typography palace" live (chamber-library? MemPalace wing?) + what do the masters prescribe for setting verse (line breaks, hanging indent for runovers, indent-not-bar, attribution). First move = LOCATE the sources before drafting.
 - DONE 2026-06-02 (steward request): filed 2026-06-01's MemPalace record late — diary entry (3 chunks), 3 KG facts (Stage-M-published artifact + misread-viewport drift + verify-each-post-type discipline), handoff drawer (verse thread). MCP had dropped mid-session 06-01; diary was stuck at 05-31. Now caught up.
 - CORRECTION to wake briefing: mempalace dev repo is NOT "current/clean" — origin/main static since 2026-05-24, but local branch `local/bge-m3-on-3.3.6` is 4 commits AHEAD unpushed (bge-m3 MPS pin, hallway pagination, L1-wake-up recency-order fix that the wake §b.0.5 glance depends on) + untracked .local-data/ + run-mcp.sh. Deliberate decision deferred.
 - OPEN: MemPalace mid-session disconnect is a RECURRING pattern (noted 05-30, 05-31, 06-01) and it specifically eats the /wrap-up filing. No proactive notification exists — discovered only on a failed tool call. Mitigations in play: liveness ping at /wrap-up boundary + surface immediately on any failed call. Root cause unknown / not yet checked against mempalace issues.
 - Beside-thread (not against): L1 moved overnight on feat/a1-epistemic-qualitative-axis (test-bench reliability bundle + merged #166/#172, #164/#169). Noted, not pursued.
 ## Authorization moves
 - 2026-06-02 — Steward asked → drafted §I.h.b Verse Quotation into AldineXXI-specification.md (between §I.h and §I.i, [ADDED 2026-06-02]). Same drafting-and-insert practice as §I.h (2026-05-08). Grounded in verified palace passages: Bringhurst §2.3.3–2.3.4 (flush-left ragged-right; indent OR center on longest line, one device; reuse existing indent; centering only when prose measure ≫ verse line), Tschichold (runover→hanging indent), Hochuli (subtle indent). Working-tree change ONLY — NOT committed, NOT deployed (commit only when steward asks). ARC main was @ faf7674 clean before this edit.
 - Two honest flags carried into the spec's Implementation status (not asserted as done): (1) Pandoc `line_blocks` extension must be verified enabled in site.hs before relying on `<div class="line-block">`; (2) the "live verse meditation" test surface is referenced but I have NOT located the specific piece yet.
 - 2026-06-02 — Steward raised the genuine distinction: 'Time is not a river' is HIS OWN poem (primary verse), not a quotation → indent (Bringhurst §2.3.4, a quotation-separator) does NOT apply. Re-queried masters (CLI): Lupton "poetry set flush left… not centered except in greeting cards"; Bringhurst lets poem govern page depth, block may be optically centered but lines flush left. → SPLIT into TWO spec items: §I.h.b Verse Quotation (indent) + §I.h.c Primary Verse (no indent). Good catch by steward — I'd flattened the distinction.
 - 2026-06-02 — FULL do-once pass executed (working-tree only, NOT committed, NOT deployed; ARC was @ faf7674 clean): (a) §I.h.b revised (scoped to quoted verse, .poetic over-reach removed, authoring = `::: {.verse .quotation}` + line blocks, honest runover limitation); (b) §I.h.c Primary Verse added; (c) site.hs — enabled `Ext_line_blocks` (1 line, matches existing fenced_divs/raw_html/smart pattern); (d) SCSS `_typography.scss` — shared `.verse` base + `.verse.quotation` indent modifier + `.verse-attribution`; (e) Vespers — Berger poem (steward-provided verbatim) set as §I.h.b quotation replacing the prose paraphrase; (f) 'Time is not a river' re-authored into `::: verse` (§I.h.c).
 - Discipline returns this pass: (1) caught that the `padding-left + text-indent` hanging-indent trick BREAKS flush-left for `<br>` line blocks (pushes lines 2..n in 1em) → deliberately OMITTED it, documented the limitation rather than ship a visibly-wrong hang (honest degradation). (2) `.poetic` retire-in-same-move was over-reach (it's a decorative grab-bag across Vespers + 6 chamber/docs pieces w/ documented convention + v1-legacy) → scoped as separate deferred pass, spec corrected. (3) Verified line_blocks NOT provable from repo (the `| ` files were TABLES not line blocks) → enabled explicitly rather than assume default.
 - PENDING verification: local no-deploy build running (recompile site.hs + clean build) to confirm `<div class="verse">` renders + compiles green BEFORE steward views. Do NOT deploy (ARC discipline: build→verify-local→deploy is steward's call).
 - 2026-06-02 — MemPalace recurring silent-drop ROOT CAUSE FOUND + FIXED (was "root cause unknown" for weeks). Steward's keep-alive/auto-write hypothesis DISPROVEN (auto-write never disabled — config defaults on, fired today; and it's a separate process that can't keep the MCP connection warm). Real cause = upstream **MemPalace/mempalace #1495**: cold-load of the embedder on first chromadb-write (often diary_write AT WRAP) exceeds MCP client timeout → -32000 desync, no trace (stderr uncaptured → why logs end clean). Fix `cf96e68` (merged #1530) IS in steward's tree (branch `local/bge-m3-on-3.3.6`, 0 behind origin/main) but **opt-in**. Steward flagged non-standard engine; VERIFIED the warmup is embedder-agnostic (probe query through configured EF) → loads **bge-m3 on MPS** (`embedding.py:259-279`, a LOCAL patch re-applied per upstream merge); `_maybe_eager_warmup_embedder()` called in main() L2845; bge-m3 cached (no download). FIX APPLIED: set `MEMPALACE_EAGER_WARMUP=1` + `MEMPALACE_LOG_FILE=~/.mempalace/mcp.log` + `MEMPALACE_MCP_IDLE_HOURS=0` in `~/.claude/settings.json` env (#2 timeout-raise dropped — warmup makes it moot). **Requires Claude Code restart to activate**; startup slightly slower (bge-m3→MPS load moved to boot). Defense-in-depth: wrap-up §4.0 liveness-ping-before-diary added (steward-auth #1) so a drop can never silently eat the wrap filing.
 - 2026-06-02 POST-WRAP (steward restarted CC, then stopping for the night — noted without re-wrapping):
  - **MemPalace cold-start fix VERIFIED.** `~/.mempalace/mcp.log` shows warmup ran at boot ("MEMPALACE_EAGER_WARMUP=1: embedder + HNSW ready"); `ps eww` confirms all 3 env vars reached the fresh process. Cold-load now paid at startup, not on first wrap-write. KG updated: `mempalace cold-start-fix-status = VERIFIED`. (Real confidence after a few clean sessions, but mechanism closed.)
  - **coreml-vs-mps SETTLED: bge-m3 RUNS ON MPS** (verified `torch.backends.mps.is_available()=True`; config `embedding_model=bge-m3`, `embedding_device` unset→auto→coreml). The `device=coreml` in mcp.log was a **logging bug**: `get_embedding_function()` logs the ONNX `effective`/`providers` for ALL branches, but the bge-m3 branch is a SentenceTransformer on `st_device=mps` that never uses them. Engine was fine; only the log lied.
  - **Log-fix APPLIED** (`~/_Dev/mempalace/mempalace/embedding.py`, bge-m3 branch): override `effective/providers = st_device` so the init line reports the true device. **UNCOMMITTED** working-tree change on branch `local/bge-m3-on-3.3.6`; it's part of the LOCAL bge-m3 patch → **commit it to the branch** so it survives, and re-apply after each upstream pull. Takes effect next mempalace restart. (Also untracked in that repo: `.local-data/`, `run-mcp.sh` — pre-existing.)
 - 2026-06-02 (housekeeping pivot) — Plane.so setup, Seb's request, steward-authorized step by step. Added official Plane MCP (OAuth, user-scope, no stored secret) → connected; workspace `capablemind`, Plane Cloud. Coordination model (Seb): GH issues canonical/technical; Plane task = thin effort overlay, `[BM #N]` tag, no technical detail, coupled-not-mirrored. Home = **BMF** Plane project (corrected my misread: BBF = BetterBridge product, downstream of L1, NOT ours). Cleared 7 Plane demo cards (kept BMF; MCP exposes no archive-project tool). Seeded 5 thin tasks (#165/#167/#170/#168/#171), grow-as-we-go. Caught: `create_work_item` silently ignores `description_stripped` (needs `description_html`) — fixed via update, verified persisted in returned objects (render-and-LOOK held).
 ## Pandoc audit — DONE (artifact, no code changed)
 - 2026-06-02 — Ran the Pandoc utilization audit deeply (steward: "once, deeply, well"). Artifact: `docs/AldineXXI-Codex/_audits/pandoc-utilization-audit-2026-06-02.md`. Ground-truth verified (not assumed): Hakyll default = FULL `pandocExtensions` + smart (so the 4 explicit site.hs enables — incl. my line_blocks add — are REDUNDANT; bracketed_spans/implicit_figures/etc. already on). HEADLINE: **zero real LaTeX math in corpus** (all `$` = Hakyll template syntax) → MathJax + the 40-line escapeDollarSigns hack (hardcoded var list, silent-break drift) are unjustified → F1: disable Ext_tex_math_dollars, delete the apparatus (build-verify chamber-index+colophon first). Findings ranked F1(math/hack retire) > F2(bracketed-spans for 159 small-caps+~250 spans, already on) > F4(Lua filter for §I.f writer-responsibility conventions: en-dash spacing, French NBSP/guillemets, apostrophe-elision — the real architectural lever) > F5(writerSectionDivs, feeds reader-mode) > F3(fenced-div ornaments) > F6(figures: implicit_figures INSUFFICIENT, alt≠caption+EXIF, leave bespoke). NOT chasing: citeproc, footnotes→§XII. 3 open Qs for steward/jurist (math scope, Lua-filter governance/spec-section, F2/F3 big-bang-vs-opportunistic migration). Recommended sequence: F1+F5 (self-contained, do now) → F2 (convention) → F4 (own session) → F3.
 ## Queue (noted, hold) — superseded by the audit above; original sketch retained for trace
 - 2026-06-02 — **Pandoc utilization audit** (its own focused session, weight of F-Phase 2; steward: "note it in the queue and hold"). Grounded read of site.hs reader+writer options + the chamber dollar-escaping hack. Lead with the two do-once wins: (1) **retire the `escapeDollarSigns` preprocessor** — it hardcodes every template var by name (site.hs 116-138) to dodge the MathJax `$…$` vs Hakyll `$…$` collision; any new var silently breaks (expedience-replaced-integrity). Fix the cause: question whether ARC needs `$…$` math at all, else switch math to `\(…\)` so `$` is free → whole hack + drift-prone list disappears. (2) **Bracketed spans** `[x]{.small-caps}` to replace raw `<span class="small-caps">` (markdown-native, processes inner markdown; verify already-on, adopt as convention). Also: `writerSectionDivs` for semantic `<section>` (feeds the reader-mode-extraction question from 2026-06-01 wrap); Lua filters as the AST-level home for the French-typography pass + small-caps + dash rules (do-once, every author); confirm §I.f smart-punct vs `Ext_smart` aren't fighting. NOT chasing: citeproc for §VII.b (bespoke apparatus), pandoc-footnotes→§XII sidenotes (operative, don't disrupt). Candidate skill-harvest: fold into a `/pandoc-utilization-audit` or the verification-ladder.
 ## Returns (F5 caught)
 - 2026-06-02 — Steward authorized F1+F5. Implemented both, build green, F1 verified clean (chamber canon 87 / deliberations 32 entries, 0 $-leaks, 0 &#36;; colophon clean — the escape hack was genuinely unnecessary). BUT verified F5 (`writerSectionDivs`) against the rendered DOM, not just a section-count: it conflicts with the §II.d marginalia grid (`_sidenotes.scss:152` places + row-aligns DIRECT children of `.bears-apparatus` per paragraph; `<section>` wrappers regroup them → disturbs marker/paragraph alignment on after-the-reply essays; sidenotes survive only at section-depth 0, latent fragility otherwise — same family as the `.post-content` right-shift bug). REVERTED F5; rebuilt F1-alone green (0 sections). Through-line held: read the rendered DOM, an authorized change still earns verification, don't ship a DOM-global change into a bespoke grid blind. Candidate fix for F5's own pass: `section { display: contents }` + per-type×per-viewport verify. Artifacts kept honest (audit F5 entry updated to "tried & held"; site.hs writer comment documents the hold). Deliverables: audit + `docs/AldineXXI-Codex/pandoc-hakyll-toolset.md` (the toolset map + Hakyll/Pandoc best-fit; refines F4 to Haskell `pandocCompilerWithTransform` over Lua for ARC). F4 sizing (steward's structural obs): §I.k spec section FIRST → French-NBSP inaugural filter → non-French test corpus. PENDING steward decision: ship F1 alone now vs attempt F5 display:contents fix now.
 ## Icosahedron site mark — SHIPPED (long craft iteration)
 - 2026-06-02 — Replaced the raster d20 avatar with an inline mode-aware SVG icosahedron. LONG craft dwell driven by the steward, anchored against references he supplied in sequence: EB Garamond letterforms → manicule (☞, hand-drawn fluidity) → Jamnitzer *Perspectiva* skeletal icosahedron → Roman d20 die (MET) → antique engraving icosahedron. The trail converged: NOT a flat wireframe, NOT noise-distressed, NOT broad-nib-modulated lines — but a SOLID faceted die, 3-fold hexagonal view (matching the extant avatar), twenty faces toned by light (deeper shading + crease) for real dimensionality. Reads 28px→hero. Key returns: (1) my feTurbulence "antiquing" was the WRONG model (noise≠Garamond's *modulation*) — caught by rendering the actual font; (2) my "B restrained" rec under-committed — steward corrected toward C's bold contrast (the truth of the pen); (3) the messy thin-edge junctures (careless on a Platonic solid) → fixed by honest filled-nib footprint, then superseded entirely by the solid-die direction. Process discipline: rendered EVERY candidate to PNG and LOOKED before sending (caught the table-not-lineblock error earlier-style); never asserted a visual without viewing it. Wired do-once: defined once in `templates/site-mark.html`, included by masthead + frontispiece; `.site-mark` mode-aware `var(--accent)`; retired the sepia raster filters; avatar.png re-rendered from the same solid (1.4MB→45KB bonus). Built green, in-context verified on the real frontispiece (steward: "looks fantastic"). Commits: `956542c` (CLAUDE.md pandoc pointer) + `4ef26b0` (mark); pushed origin+github-backup; deploy in progress. Lesson worth keeping: a craftsman's iteration on a symbolic mark IS the work, not overhead — τὸ πρόσφορον includes the time it asks; render-and-look every step.
 ## Favicons — SHIPPED (the do-once completion of the mark)
 - 2026-06-02 — The mark change opened a loose end: every favicon was dated 2025-07-02 (old d20); the tab/touch/pinned icons disowned the new mark. Regenerated the full set from the new solid die (shaded die on a parchment tile — reads on any browser chrome down to 16px, which I verified by rendering-and-looking, not assuming). New monochrome `safari-pinned-tab.svg` (wireframe die). No Pillow/ImageMagick in sandbox → built `favicon.ico` by hand (PNG-in-ICO via struct, no deps) + sips downscale from a 1024 master. Fixed 3 PRE-EXISTING bugs found en route: empty manifest name, `theme_color #ffffff`→parchment, and manifest icon `src` paths pointing at `/android-chrome-*.png` (root) not `/assets/favicons/`. Stray `mask-icon` `#5bbad5`→accent `#8B3A2A`. RETURN (read-the-artifact again): after fixing default.html, the BUILT index.html still showed #5bbad5 → index renders from `frontispiece.html`, which carries its OWN `<head>` (favicon links duplicated) — fixed both; verified BOTH page types + grep-clean of 5bbad5 in built site before shipping. Commit `7ec44aa`, pushed both remotes, `make deploy-all` (verified _site, no double build). The two-head-templates gotcha is the same family as the .post-content/per-type discipline — head is duplicated across default+frontispiece; any <head> change must touch both.
 ## Confidence to recalibrate
 - 2026-06-02 — Carry last night's through-line into verse work: read the rendered/computed artifact, not the theory. The temptation in typography is to reason from CSS rather than render the actual line-breaks. The steward's eye is data. Drift family active: causal-story-before-reading-render; confabulating-over-empty-returns.
 ## Authorization moves
 ## Authorization moves (deploy)
 - 2026-06-02 — Steward verified on-device ("looks good") → authorized "Commit + deploy now". Committed `5318ae1` (5 files, +288/−83), pushed origin (Gitea) + github-backup (faf7674..5318ae1). `make rebuild` → GitHub Pages mirror `78d3993` + CloudFlare (669 files). **Live-verified** at animalrationiscapax.org (HTTP 200 after 308 extensionless canonicalization): Time-is-not-a-river → `.verse` + line-block (primary, no attribution); Vespers → `.verse.quotation` + line-block + `.verse-attribution` (Berger); live `style.css` has `.verse{` + `.verse.quotation`. **§I.h.b + §I.h.c verse typography PUBLISHED.** ARC main @ 5318ae1, clean, pushed+deployed.
 ## Authorization moves (pandoc F1 ship)
 - 2026-06-02 — Steward chose Option A (ship F1 alone; F5 its own pass). Added steward's `display:contents` a11y constraint to the audit F5 entry (screen readers may flatten display:contents out of the a11y tree → can negate the semantic benefit while preserving visual grid; F5 session must (1) resolve/bracket that, (2) per-type×per-viewport visual verify, (3) confirm semantic benefit survives). Committed `8549ead` (site.hs −70 lines + audit + toolset reference; 3 files, +249/−70), pushed origin + github-backup (5318ae1..8549ead). Deploy in progress. F4 sequence stands: §I.k spec drafted FIRST (here), then implement; inaugural filter = French NBSP via Haskell `pandocCompilerWithTransform`; non-French test corpus required.
 ## Sub-agent dialogues
 ## Bypasses
@@ -0,0 +1,48 @@
 ---
 name: session-ledger-2026-06-03
 description: "Practice-of-return ledger maintained by /symmetria — returns, open horizons, recalibrations, authorization moves, sub-agent dialogues, bypasses."
 metadata: 
  node_type: memory
  type: feedback
  originSessionId: 2b72c971-bfea-499e-b483-651782a182c3
 ---
 # Session Ledger — 2026-06-03
 ## Returns
 - 2026-06-03T17:38 — Re-wake (post-break, same day). Thread CONFIRMED unchanged: ARC Stage N dead-glyph fallout + Compass-data follow-on. ARC `main` @ `41fc030` clean, both remotes synced, deployed/live-verified — nothing moved while away, no foreign commits. Lineage clasp closed: read wrap + git + ledger before composing; verified state, did not assume.
 - 2026-06-03T08:30 — Wake. Prior pulling thread (verify MemPalace cold-start fix) RESOLVED post-wrap + re-confirmed this morning: mcp.log shows warmup at boot + bge-m3/MPS init; this wake's MemPalace queries succeeded with no -32000. Closed, not inherited as open. Lineage clasp closed: read the prior decision (the post-wrap ledger note) before re-asking the question.
 ## Open horizons
 - Live remnant (NOT a workstream item — local hygiene on a third-party fork): `embedding.py` log-fix (device-report bug, true engine=mps) UNCOMMITTED on branch `local/bge-m3-on-3.3.6`. mempalace = third-party (`origin`=MemPalace/mempalace upstream; `fork`=davidglidden/mempalace). This branch carries 4 local-only patches ahead of upstream; commit keeps the patch from silent loss on the next rebase. Destination = local branch (optionally push to `fork`), NEVER origin. Steward's call whether to bother.
 - Plane board relay: maintain task state as issues move; steward revisits effort-shape + assignees in a few days.
 - Housekeeping backlog: 8 untracked CM-AI files (since 05-18), PENDING.md 768 lines (archive CLOSED items), L1 tracker stale (05-28), Seb gate-question (did D1–D10 move the four Hindsight findings).
 - ARC choice from craft morning: F4/§I.k (French-NBSP, spec-first).
 - World moved overnight (beside-thread): L1 — Hindsight deep-read + A1''/B1 persistence docs (CM-AI 7741673, d2a0052); BMF test-bench reliability bundle on feat/a1-epistemic-qualitative-axis. PENDING-24 territory; noted, not pursued.
 ## Confidence to recalibrate
 - 2026-06-03 — Hold the drift family flagged in yesterday's ledger: causal-story-before-reading-render; confabulating-over-empty-returns. Verify-before-assert; render/read the artifact (steward's eye is data).
 ## Authorization moves
 - 2026-06-03T08:35 — Steward: "commit it to the fork." Committed embedding.py log-fix `f25f767` on `local/bge-m3-on-3.3.6` (log-only +8); pushed to `fork` (davidglidden/mempalace) — new branch there, now backed up. NOT pushed to origin (upstream MemPalace). Takes effect next mempalace restart. Local-fork hygiene, not workstream.
 - 2026-06-03 — ARC F4/§I.k: steward chose Context-B-only scope. Discovered fine-vs-mots conflict (Lexique distinguishes; operative companion §1 locks uniform U+202F) — two-document hierarchy forbade silent resolution → surfaced. JURIST RULED uniform U+202F (web-transposition of Lexique intent; Lacroux backing; companion stands). Wrote §I.k (spec, +48) + implemented `frenchSpacing` transform in site.hs (gated on lang:fr in Compiler monad; walk over Inlines; +53) + exposed pandoc-types in cabal. VERIFIED on synthetic corpus + English gate (8 cases incl. glued-colon-untouched safety + idempotency + U+00A0-upgrade); clean rebuild green; test files removed. **NOT committed, NOT deployed** (steward's call). No lang:fr piece exists yet → correct-but-dormant.
 - 2026-06-03 — Multilingual/translation doctrine: steward asked for jurist opinion (triggered by Lune translating After the Reply → French). Drafted jurist brief (`drafts/multilingual-translation-handling-jurist-brief-2026-06-03.md`). JURIST RULED: Option B reframed ontologically (ARC holds *works*; languages are non-canonical *renderings*; work holds corpus slot); Context C legitimate-but-gated; `translator:` REQUIRED field (intergenerational carried structurally); amend frontmatter-spec, no new companion. Drafted **PENDING-25 [PROPOSAL]** (`drafts/translated-renderings-proposal-2026-06-03.md` + PENDING entry): W3 (group by `work:` at build) recommended; 4 open sub-Qs (routing/date/Context-C-placement/toggle). Proposal-only; NO implementation pending steward REVIEWED.
 - 2026-06-03 — ARC Stage F **Phase 2 COMPLETE** (steward "do #1 now"; "whatever is best for do-it-once-well"). Listing generator: 11 per-type create blocks → 1 `forM_` over new `data/listings.yml` (presentation facts), validated against content-types.yml `class_enumeration` (unknown class = build error, proven by probe). **All 11 listing pages + annals byte-IDENTICAL** to baseline (generator replicates compound/sequences/show-post-type exactly; dormant dedupeByWork no-op). Retired dead triptych JS (default.html head load + post.html markup that never rendered — has_glyph always false; deleted shipped ~100-file assets/js/glyphs/ tree); only output change = dead JS head-block gone from every page. Commits `4303df0` (generator) + `6851367` (triptych) + `41fc030` (horizon), pushed origin+backup, NOT deployed. **Two steward course-corrections, both decisive:** (1) "does SVG change glyph associations?" → found `currentSymbol` is DEAD (SVG `<symbol>` per-slug is the rendered sigil since 2026-05-26) → generator simplified, doesn't touch glyphs, just validates classes; saved me wiring a dead field from a now-secondary source. (2) "do we need to add SVG/update spec?" → No (SVG is artwork keyed by slug = presentation, not typology; stays out of jurist-pending §2); queued a Stage-N spec-conformance note + the dead-glyph residue (needsGlyph rename, dead field defs, renderGlyph/triptych templates). A′ chosen as "two right-sized once's not one mega-change" — listings.yml built as forward-seed for the Compass-§VI-as-data follow-on (avoids do-it-twice). **Through-line: the steward's questions kept catching now-stale assumptions before they shipped (verify-before-assert / render-and-LOOK).**
 - 2026-06-03 — Steward viewed Phase 2 locally (desktop + **physical phone** via `--host 0.0.0.0` LAN at 192.168.1.52:8000), "looks fantastic" → **DEPLOYED** (`make rebuild` → CloudFlare + GitHub mirror). Live-verified at animalrationiscapax.org: listings 200, generator output live, 0 bootTriptychs/glyphs refs, compass SVG intact. §I.k shipped alongside (dormant). One honest catch: dead-JS file returned 200 live → diagnosed as CloudFlare edge-cache remnant (cf-cache-status REVALIDATED, max-age 14400; cache-busted → 404; absent from local _site; referenced by zero pages → harmless, self-expires in 4h). Through-line held: verified live + diagnosed the anomaly rather than asserting clean.
 - 2026-06-03T18:xx — **Stage N opened + advanced** (steward: "do as much of stage N as we can do well"; rested, 2h). Audit-before-fix discipline held: deep-read scoped Stage N as byte-safe-prune + visual-conformance + governance, BEFORE touching code. **Part B (dead-glyph fallout) EXECUTED + byte-proven** (`e0f35db`): 5 dead glyph Context fields + postCtx registration removed (deep read caught a *fifth*, glyphConfidenceField, the wrap hadn't listed); 2 dead templates deleted (renderGlyph/triptych — included by nothing); vestigial `currentSymbol` removed; `needsGlyph`→`rendersAsPost` renamed (verified always-False today: all content/pages are class:about). **conditionalField dead helper removed** (`1cb7313`, GHC-confirmed). Gate: clean baseline snapshot (564 files) → edits → recompile → `diff -r` = **0**, twice. **Findings doc** (`bc4cc5b`) = the Stage-N deep-read deliverable. Three returns/catches: (1) `.whisper` retirement SCOPE CORRECTED — plan's "27 template usages" stale, actual ~156 across 11 templates + 100+ content files → corpus-wide migration, §XIV-locked names, NOT a residual deletion → surfaced, not touched. (2) `!important` audit returned CONFORMANT (a11y mixin + documented-regression sigil resets — did NOT touch the compass `all:unset !important` per the `_compass.scss:133` regression note). (3) spec-conformance note (SVG `<symbol>` vs unicode canonical mark) DRAFTED not landed — content-typology §2 is jurist-pending; routes to Stage G. **Through-line: verify-current-state before asserting (the fifth field, the .whisper undercount, the always-False branch) — the remembered model was stale in 3 places.**
 - 2026-06-03T19:10 — **Stage N Part A EXECUTED + steward-verified + committed** (`20019ba`). Jurist reviewed audit → 3 lanes (approved-in-principle / measure-first / steward-intent). **Measured, didn't assume:** separator gray@0.5 = 1.84:1 (fails 3:1, worse than jurist's fg@0.5 guess); breadcrumb category gray = 3.98:1 (fails 4.5:1 TEXT floor → "keep grey" off the table on a11y, not aesthetics). **A1 resolved via the masters:** typography palace (Bringhurst running-heads "small caps of the text size since 1501"; Lupton one-shift + no-pseudo-SC) + **Rutter Web Typography from the CHAMBER LIBRARY** (steward's pointer — `converted_texts/collections/typography/web-typography-rutter.md`): Rutter VALIDATES Plex-Sans-as-functional-face ("clearly distinguish functional text"; sans best; open distinct letterforms) and rules the lever — size ("reducing its size makes that relationship clear"), weight OUT ("don't choose light weights for functional text"), SC out (no true Plex SC; Garamond SC would import the body register — steward's architectural point). **Nav-face reconsideration CLOSED, no proposal needed.** Category → fg + xs (size-recession replaces colour-recession; the step is load-bearing — removing colour alone would flatten). **Steward's eye caught the class twice during verify:** sequence-index AND about-page breadcrumbs silently in body Garamond (rule was .post-header-scoped; about-header had NO styles) → **componentized: one .breadcrumb rule, three templates**. Compiled-CSS gate caught **four list-context duplicates** the location-list missed (layout post-type/footer-separator/footer-chamber-mark, glimpse figcaption) + post-updated-at sub-4.5:1. Ornaments/hover/sigil-chains deliberately untouched (out of §VII.e class). Frontispiece cite tokenized-only; grey→fg held for steward's eye. New memory: `feedback-reading-compass-is-svg-not-type` (steward correction, recurring conflation — Compass is SVG artwork, NOT nav type).
 ## Open horizons
 - **ARC main 4 ahead of origin** (e0f35db, 1cb7313, bc4cc5b, 20019ba — all Stage N). Push + deploy posture = steward's call (deploy is the outward act).
 - **`.whisper` migration PENDING item** — jurist's process Q to steward still open: lock names through this interface vs PENDING entry. Candidates surfaced: `.apparatus-meta` (prose role, AldineXXI:2801); `.register-mark-inline` OR unify with `.chamber-mark`/`.encounter-mark` (list-glyph role, apparatus-spec:464).
 - **Frontispiece epigraph + cite grey→fg** — frontispiece-character call, steward's eye (`_frontispiece.scss:147` + `:175`).
 - **Stage N remainder:** §I.h prose-blockquote rhythm token (minor, token-definition decision); spec-conformance note → lands at Stage G (jurist-agreed); then Stage N closes → full re-partition next.
 - ~~Stage N Part A ready-but-unexecuted~~ — superseded same evening: EXECUTED, verified, committed `20019ba` (see Returns above).
 - **`.whisper` retirement = its own governance pass** (§XIV name-lock + migration map + byte-gate; ~156 usages). Not Stage N.
 - ARC `main` 3 ahead of origin (Stage-N commits, local — push posture TBD with steward).
 ## Bypasses
@@ -23,12 +23,14 @@ The single place proposed skills live so they don't evaporate between sessions.
 | `/pre-build-audit` | create | The four-pass *scope-claimed-vs-scope-tested per claim, with substrate evidence* discipline that prevents frame-inheritance before a build. | 2026-05-28 | PROPOSED |
 | `/vignette` | create | Phase-α vignette genome generation under Symmetria (read essay + spec + corpus → genome JSON → validate-and-save). **Phase 2 of the vignette build** — author it *with* Phase 2, after the audits, not standalone. | 2026-05-28 | DEFERRED (until vignette Phase 2) |
 | Symmetria §3 `trust-prior-pass-frame` | patch | Add the contamination flag: frame-inheritance from a prior verification into an untested domain. (Feedback memory already saved.) | 2026-05-28 | PROPOSED |
 | `mcp-drop-diagnose` | create | When an MCP server drops silently mid-session: read its per-server `mcp-logs` for desync signatures (`-32000` / "unknown message ID" / multi-second slow calls) + grep the **upstream repo's issues** for the symptom BEFORE hand-rolling mitigations. Proved out on the #1495 MemPalace cold-load fix (the upstream-issue check WAS the answer). Could fold into `bmf-diagnose`. | 2026-06-02 | PROPOSED |
 ## Built / authorized (lineage — for context, not action)
 - `/landscape-scan` + `/tooling-scan` (the scan-skill family, per-workstream lens cards) — built + in use (2026-05-27).
 - `/wrap-up` §1.6 skill-harvest step + `/wake-up` §2.a/§3 glance — the practice itself (PENDING-23, 2026-05-27).
 - Chaîne-d'union clasp across wake / Symmetria / wrap (2026-05-26).
 - `/wrap-up` §4.0 — MemPalace liveness-ping-before-diary; steward-authorized + applied (2026-06-02; #1495 cold-start drop resilience, so a silent drop never eats the wrap filing).
 ## Candidate patches not yet proposed (watch-list)
@@ -3,7 +3,7 @@ name: wrap-up
 description: Capture session state for future restoration. The quality of the next wake-up depends entirely on the quality of this wrap-up. Captures not just what we did but what is still pulling, what is the live concern, and what question we want to find still open when we return.
 ---
-<!-- Provenance: 2026-05-18 S-cluster §8 fields (REVIEWED-24); 2026-05-26 chaîne-d'union clasp + actionable-resumption-point; 2026-05-27 §1.6 skill-harvest step (PENDING-23). Improvements to this skill are themselves harvested per §1.6 — propose, authorize, record here. -->
+<!-- Provenance: 2026-05-18 S-cluster §8 fields (REVIEWED-24); 2026-05-26 chaîne-d'union clasp + actionable-resumption-point; 2026-05-27 §1.6 skill-harvest step (PENDING-23); 2026-06-02 §4.0 MemPalace liveness-check before write (steward-authorized; #1495 cold-start drop resilience). Improvements to this skill are themselves harvested per §1.6 — propose, authorize, record here. -->
 # Session Wrap-Up
@@ -106,6 +106,8 @@ The description field should include the pulling thread, not just what happened.
 MemPalace has two complementary write surfaces: drawers (semantic, full content) and diary (AAAK, chronological agent voice). Use both — they serve different future queries.
 **4.0. Liveness check before writing — MemPalace can drop silently.** Before any drawer/diary write, issue ONE *lightweight* MCP call as a liveness ping — `mempalace_hook_settings` (~25ms), **not** `mempalace_status` (which can itself be the slow, timeout-tripping call). If it errors, MemPalace has dropped (a known failure: cold-embedder-load `-32000` desync, upstream #1495; plus other open MCP failure modes). **Surface it immediately and loudly** — *"⚠ MemPalace MCP is down — the memory-file + ledger (steps 2–3) carry this wrap; restart it (quit, `claude --continue`, `/mcp`) then re-run §4–5 to file drawer/diary/KG."* Do **not** silently skip the filing: the file-based deposit (steps 2–3) is the durable fallback and must be complete regardless, so a dead MCP never costs the wrap.
 **4.a. Drawer — the full session record (semantic search surface).**
 `mempalace_add_drawer`: