🔧 Auto-commit from sysupdate on 2026-06-02 09:21

2026-06-02 09:21:11 +02:00
parent fffcf173c1
commit 29b3d43970
15 changed files with 688 additions and 1 deletions
@@ -2,18 +2,32 @@
 ## Standing preferences
 - [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.
 - [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`.
 ## 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`.*
- [ARC](project-arc-rework.md) — canonical workstream tracker; chronological accumulating record 2026-04-16 →. Companion canonical references in the ARC repo: `docs/AldineXXI-Codex/_audits/spec-corpus-audit-2026-05-16.md`, `arc-state-survey-2026-05-16.md`, `audit-delta-2026-05-19.md`; `docs/decisions/2026-05-16-type-applicability-matrix.md` (ADR-006 v2). **2026-05-25 — SECOND EDITION PUBLISHED.** Stage F dwellings/Compass live on `animalrationiscapax.org` (13-point Compass, `/dwellings`). **2026-05-26 — glyph optical-sizing → SVG migration shipped live** (`e1d0f12`); Gloss ✎→¶; pairing recast annotator/reader. **2026-05-28 (morning)** — Stinkkatze fragment + font-backup cleanup (`d6f5cdb`). **2026-05-28 (evening) — content-types.yml generator landed** (`a9dc188`; script + YAML + Makefile + CLAUDE.md, ~350 LOC + 470 LOC YAML; `data/content-types.yml` carries 14 content types + Annals view + Chamber seal + class_enumeration; `make content-types` + `make content-types-check`; loud-on-broken-matrix discipline; preserves surface forms verbatim per ADR-006). **Then: vignette spec revision brief drafted** (`438ee03`; `docs/AldineXXI-Codex/drafts/vignette-spec-revision-brief-2026-05-28.md`, 270 lines, 9 sections) per steward's Shape C reframing (Claude Code as Phase-α engine, /vignette skill as codified discipline, interface contract as transition mechanism, system-prompt-as-artifact dissolves; CapableMind+local arrives Phase-ω). Stage F remaining: per-piece marginalia render + the audits + (minor) a11y patch + **Phase 1 vignette renderer build tomorrow** (engine-independent; can proceed against hand-written test genome in parallel with jurist's spec revision). Stage G (jurist → content-typology OPERATIVE) closes. ARC on `main` @ `438ee03`, 2 ahead of origin (local-only by directive). Tracker file updated through 2026-05-26; needs append for 2026-05-28 evening at next session.
+- [ARC](project-arc-rework.md) — canonical workstream tracker; chronological accumulating record 2026-04-16 →. Companion canonical references in the ARC repo: `docs/AldineXXI-Codex/_audits/spec-corpus-audit-2026-05-16.md`, `arc-state-survey-2026-05-16.md`, `audit-delta-2026-05-19.md`; `docs/decisions/2026-05-16-type-applicability-matrix.md` (ADR-006 v2). **2026-05-25 — SECOND EDITION PUBLISHED.** Stage F dwellings/Compass live on `animalrationiscapax.org` (13-point Compass, `/dwellings`). **2026-05-26 — glyph optical-sizing → SVG migration shipped live** (`e1d0f12`); Gloss ✎→¶; pairing recast annotator/reader. **2026-05-28 (morning)** — Stinkkatze fragment + font-backup cleanup (`d6f5cdb`). **2026-05-28 (evening) — content-types.yml generator landed** (`a9dc188`; script + YAML + Makefile + CLAUDE.md, ~350 LOC + 470 LOC YAML; `data/content-types.yml` carries 14 content types + Annals view + Chamber seal + class_enumeration; `make content-types` + `make content-types-check`; loud-on-broken-matrix discipline; preserves surface forms verbatim per ADR-006). **Then: vignette spec revision brief drafted** (`438ee03`; `docs/AldineXXI-Codex/drafts/vignette-spec-revision-brief-2026-05-28.md`, 270 lines, 9 sections) per steward's Shape C reframing (Claude Code as Phase-α engine, /vignette skill as codified discipline, interface contract as transition mechanism, system-prompt-as-artifact dissolves; CapableMind+local arrives Phase-ω). Stage F remaining: per-piece marginalia render + the audits + (minor) a11y patch + **Phase 1 vignette renderer build tomorrow** (engine-independent; can proceed against hand-written test genome in parallel with jurist's spec revision). Stage G (jurist → content-typology OPERATIVE) closes. ARC on `main` @ `438ee03`, 2 ahead of origin (local-only by directive). Tracker file updated through 2026-05-26; needs append for 2026-05-28-evening + the Stage-M arc. **2026-05-29→31 — Stage-F audits + Phase-1 FIX items + §II.d marginalia `:has()` gate + glimpse full-bleed fix; mode modernization pulled forward as Stage M.** **2026-06-01 — STAGE M COMPLETE + PUBLISHED LIVE** (`faf7674`; deployed CloudFlare + GitHub mirror; verified at `animalrationiscapax.org`): system light/dark via `light-dark()`+`color-scheme`, §I.h block quote, toggle+JS removed, P3 wide-gamut, prefers-contrast — plus **`.post-content` purge** (fixed the marginalia-post right-shift: a non-spec Jekyll wrapper double-matched the `:has()` grid → nested grids; about-pages were fine because they have no wrapper) + **`.callout` retire** (spec residual). Wide-window stance accepted (grid needs ~66–70rem to centre). ARC `main` clean, 0 unpushed. **Next: §I.h.b Verse Quotation (read the masters — Bringhurst — first), icosahedron→accent SVG, F-Phase 2 (Hakyll listing generator), §VII.e residual.**
 - Chamber-typography — *tracker not yet established*; substantive moves live in per-session memories (2026-05-11 onward) + `project-chamber-cruft-restoration.md` + `project-chamber-typography-mining-plan-2026-05-15.md`.
 - [ARC chamber v1-legacy cluster](project-arc-chamber-v1-legacy-cluster.md) — ARC's `content/chamber/**` is intentional v1-Chamber legacy (not drift); deferred to-do = gather into a presentable cluster as a record of development. Out of §5 clause-1 audit scope. Confirmed 2026-05-29.
 - Studium Engine — *tracker not yet established*; substantive moves live in per-session memories (2026-05-15 onward) + the seed brief at `~/_Dev/studium-engine/`.
 - [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 2026-06-01 — 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)
 - [Session 2026-05-31 — ARC Stage-F Phase-1 FIX items landed + mode modernization pulled forward (Stage M)](session-2026-05-31-arc-phase1-fix-items-mode-modernization-pulled-forward.md) — **Closed all Phase-1 FIX items** then took a sequencing decision. **5 commits, ARC `main` ahead 24, local-only, clean, nothing deployed.** Committed (all verified; gate+glimpse+compass-a11y confirmed on the steward's **physical phone**, a11y via **VoiceOver**): `b228f47` stray `.bak` removal · `913aaea` **§II.d `:has()` marginalia gate** (no-marginalia → symmetric body centred at **38rem** per spec L260, not 720px) · `f390b5d` **glimpse phone full-bleed fix** (PRE-EXISTING LIVE bug; root cause = TWO clips, parent `<figure>` overflow-x:auto AND `.wrapper/.page-content/main` overflow-x:hidden, both must open; **took 3 attempts — 2 wrong diagnoses before reading the full ancestor chain**) · `02393bc` **compass §3F+§3G** (dead `.path-*` `@extend` block removed — all 7 inert Jekyll-era; a11y = visually-hidden text node over `aria-label` for durability: translatable/CSS-off-survivable/native) · `94b359e` **plan amendment — mode modernization PULLED FORWARD as Stage M**. **§I.h blockquote DEFERRED to ride Stage M** (Option B; clean form needs `_dark` gone first). **THE DECISION: mode modernization pulled forward (Stage M, beside F)** — resolves the closing plan's self-contradiction (re-partition couldn't both close F and follow an F+1 step). Full re-partition now follows M. Amended all 3 records in step (anti-drift): plan §5c+§6, ADR-006 Part 5, this tracker. **Process:** tool channel flaked → **confabulated unread content over empty returns twice** + glimpse double-mis-diagnosis; caught each; through-line = **read the rendered/built artifact, an empty tool return is not a read**. **Pulling thread: Stage M(a) — the `light-dark()` structural core — fresh next session** (steward's call, for energy). **Literal Q:** inventory the 7 scattered dark blocks' non-flip choices FIRST, or migrate `main.scss` tokens first? + M(a) one commit or split? **Standing flag:** glimpse fix corrects a LIVE bug, committed-not-deployed. Skill harvest: `render-chain/ancestor-clip` diagnostic + SCSS verification-ladder (reinforced); carry the prior unauth'd set.
 - [Session 2026-05-30 — ARC Stage F: both audits run, 3 reconciles, RETIRE+DEDUPE, mode-mechanism modernization decided](session-2026-05-30-arc-stage-f-audits-reconciles-retire-dedupe-mode-mechanism.md) — **Ran both Stage-F audits** (Hakyll feasibility CONFIRMED from unpacked 4.16.6.0 source — Rules monad + preprocess; SCSS gap-map = cut-by-page-area-vs-scale, cleanup-first then conditional re-partition). **3 reconciles:** Compass surfacing (dream/heteronym/hommage/reverie→`surfaced`; reverie ruled surfaced; YAML regen); **the ADR Stage-E record fiasco** (E was merged but ADR Part5 frozen at authoring → dated stage-block + **standing anti-drift rule: ADR-006 Part5 canonical, `project-arc-rework.md` mirrors, update both at stage-close**; my drift = anchored on stale audit-delta snapshot); user-visible **titleClasses bug** (6 pieces rendered title-less vs §2 `title_display:shown` → stopgap fix, durable=table-driven). **Stage-F-closing plan** (`drafts/stage-f-closing-plan-2026-05-30.md`; two tracks SCSS+Hakyll; **vignette→F+1**; **marginalia=SCSS via `:has()`**). **Phase 0 RETIRE** (~190 dead SCSS lines — voice colours/$enable/metadata/15 mixins; **byte-identical CSS**) + **Phase 1 DEDUPE** (blockquote/enfilade/small-caps, per-selector-diff verified) + small-caps drift fix (one small-caps, SC face) + token rename `$sidenote-size`→`$marginalia-register-size`. **THE BIG ONE — mode-mechanism modernization DECIDED** (the thing the steward had circled): drop the toggle, **adopt the system** via `color-scheme`+`light-dark()` (pure CSS, zero JS), **+P3 +prefers-contrast**; principle = *honor the reader's environment, never ask them to re-declare it* (font-size analogy generalized); founds the vignette; **SHELVES the dark→var MIGRATE** (superseded — light-dark deletes `_dark`+scattered blocks wholesale; caught before investing). Note `drafts/mode-mechanism-modernization-2026-05-30.md`. **10 commits; ARC `main` 19 ahead, local-only, working tree clean; nothing deployed.** **Pulling thread: Phase 1 FIX** (`:has()` marginalia-render gate + §I.h blockquote fix + stale compass `@extend`s + compass a11y) → re-read gate → Stage F closes; all theme-independent (mode rework shelved to F+1). **Literal Q:** which FIX leads — marginalia `:has()` gate or §I.h blockquote — and does the steward want to *see* the §I.h blockquote render (border + forced-italic both gone) before commit? **MemPalace MCP disconnected mid-session — drawer/diary/KG NOT filed; memory file + ledger carry the record (re-file if reconnected).** Skill harvest: propose **SCSS-refactor render-equivalence verification-ladder**; carry `/spec-code-audit` + `bmf-diagnose` + `/arc-typeset` + `/pre-build-audit` + `/vignette` + `trust-prior-pass-frame` (+ new `anchoring-on-stale-snapshot`).
 ## Archived (2026-05-29 — vignette spec landed + code-represents-spec audit + Hakyll north-star; the thread today woke from)
 - [Session 2026-05-29 — ARC: vignette spec landed (jurist pass) + code-represents-spec audit + Hakyll north-star](session-2026-05-29-arc-vignette-spec-landed-code-audit-hakyll-northstar.md) — **Woke into vignette Phase 1; became the Stage-F-closing groundwork.** Settled vignette open-Qs → **jurist revision landed canonical** (`c52fb69`; jurist sharpened dwell-test→mandatory-first-pass, transition-trigger→+L1-precondition, surfaced the Phase-1 **orientation-preserving-luminance trap** — a dark-mode bug that hides from visual review). Steward reframed: **audits before the vignette build**. Ran **code-represents-spec audit** (§5 spine, `b15684a`) — §5 largely honored, **2 gaps**: clause-2 glyph drift FIXED (`164d8c3` ✎→¶), clause-3 silent fallback ships-with-Stage-G; no spec-gaps → **Stage G unblocked**. `content/chamber/**` = intentional v1-legacy (captured `project-arc-chamber-v1-legacy-cluster.md`). Wrote **Hakyll north-star** (`d714a11`; build *reads* content-types.yml as single source, add-type = YAML-row; closes clauses 2/3 *durably*). CLAUDE.md Hakyll 4.16.2.1→**4.16.6.0** (`7bc2a8a`). **Pulling thread: run the Hakyll-simplicity + SCSS-cascade audits tomorrow — the gate to the plan that completes composition, writes+tests the vignette, and closes Stage F.** **Literal Q:** which audit leads (Hakyll or SCSS) + does Hakyll 4.16.6.0's `Rules` monad confirm the data-driven generation the north-star bets on (confirm at step 0 via `stack unpack`)? ARC `main` @ `7bc2a8a`, **8 ahead of origin** (local-only). Skill harvest: propose **`/spec-code-audit`**; carry bmf-diagnose + /arc-typeset + /pre-build-audit + /vignette + trust-prior-pass-frame.
 ## Archived (2026-05-28 evening — ARC Stage F generator + vignette scoping; the thread this session woke from)
 - [Session 2026-05-28 (evening) — ARC Stage F: content-types.yml generator + vignette spec revision brief (Shape C reframing)](session-2026-05-28-evening-arc-stage-f-vignette-scoping.md) — **Woke 36s after L1 wrap (collapsed pause — context reset within same evening, not anticipated overnight wake); steward pivoted to ARC.** Two clean moves: (1) **content-types.yml generator landed** (`a9dc188`) — chose by steward over lighter Stage F alternatives because it's Stage G's gate; composed via dwell-before-composing (5 shape-decisions confirmed before writing a line: data path / class enum inclusion / surface-form fidelity / italic stripping / --check); script + YAML + Makefile + CLAUDE.md atomically, ~350 LOC + 470 YAML; verified loud-on-broken-matrix + --check honest in both directions + glyphs round-trip (◧ Photo Essay, ¶ Gloss, ☞︎ Readings with VS, ⟐ Chamber); Annals `class: null` under `views.annals`, Chamber `class: null` under `apparatus_seals.chamber` with `inherits_host: true`. (2) **Vignette engine scoping → revision brief drafted** (`438ee03`) — re-read 334-line spec end-to-end (trust-prior-pass-frame discipline held); surfaced state-of-world (clean greenfield: zero sidecars, no template refs, no SCSS impl beyond placeholders; old triptych at 31k LOC dormant since site.hs:1259 DISABLED note); 3-component / 4-phase architecture; risk on Phase 1 renderer; 5 under-specified items (dark mode, reduced motion, cycle-end, FCP, ground-mark corpus-relative weight). **Steward's three reframings reshaped the work**: (a) "spec needs revisit before building"; (b) "local LLM not possible yet — Claude until CapableMind matures"; (c) **"I'd rather *you* be the engine"** → **Shape C** (category-shift from Shape A/B framing) — Claude Code as Phase-α engine; `/vignette` skill as codified discipline replacing system-prompt-as-artifact (which dissolves); interface contract elevated to load-bearing transition mechanism; Symmetria-per-generation as Phase-α governance equivalent. Brief filed at `docs/AldineXXI-Codex/drafts/vignette-spec-revision-brief-2026-05-28.md`, 270 lines, 9 sections (preserved / reversed / elevated / addition / tightening / risk reassessment / open questions for jurist / sequencing / not-this); 7 jurist-judgment questions ranked. **Phase-α / Phase-ω naming used throughout** (bridge metaphor; "we will cross that bridge"). **Returns**: dwell-before-composing paid off twice (generator + scoping); trust-prior-pass-frame held through scoping; recognized Shape C as category-shift not Shape A/B variant. **Recalibrations**: no mid-session Symmetria pulses (L1→ARC pivot + Shape A/B→C reframing were both pulse-worthy); sequencing assumption drifted optimistic on jurist tomorrow availability (brief §8 has fallback). **Pulling thread now: build Phase 1 of vignette renderer tomorrow** (engine-independent prototype against hand-written test genome; can proceed in parallel with jurist revision; only Phase 2 — actual /vignette skill + harness — waits for revised spec). L1 Monday-package thread sits beside (not behind) — both can be worked tomorrow. **Literal Q**: did the steward read the brief overnight, what does he want to refine before Phase 1 begins — or proceed straight to prototype trusting the brief's executor-leans? Secondary: any movement on the L1 Monday-package thread? **Pause**: brief is the chain for tomorrow's build (or for jurist revision first; steward's call); Phase 1 doesn't strictly need jurist pass first because renderer is engine-independent. Skill harvest carrying: `bmf-diagnose` + `/arc-typeset` + `/pre-build-audit` + Symmetria `trust-prior-pass-frame` flag (all from morning); **new this session: `/vignette` skill** (named in brief; authoring is Phase 2 work under Shape C). ARC `main` @ `438ee03`, 2 ahead of origin (local-only by directive).
 ## Archived (2026-05-28 — L1 pre-build audit + Monday package; the morning the evening woke from)
@@ -0,0 +1,16 @@
 ---
 name: mempalace-kg-object-128-char-cap
 description: "MemPalace kg_add `object` field hard-caps at 128 chars — write KG objects short on the FIRST pass; put detail in the drawer."
 metadata: 
  node_type: memory
  type: feedback
  originSessionId: 17ef4cde-4d62-41e8-8098-52898f013b43
 ---
 `mcp__mempalace__mempalace_kg_add` rejects any `object` longer than **128 characters** (`"object exceeds maximum length of 128 characters"`). This recurs at nearly every `/wrap-up` §5 — I write a rich, sentence-length object, it fails, I re-add concise. The steward flagged it 2026-05-31: *"this happens every time."*
 **Why:** the KG is a graph foothold for retrieval, not the record. The full session detail belongs in the **drawer** (`add_drawer`, no such cap) and the **session memory file**. The triple just needs to be findable by `kg_query` on the subject.
 **How to apply:** at `/wrap-up` §5, compose every `kg_add` `object` as a **short phrase under 128 chars on the first call** — keyword-dense, not a sentence. Park the elaboration (commits, rationale, caveats) in the drawer + memory file, and if a pointer is useful, end the object with `…; see <doc> §X`. Subject/predicate also stay terse. One write, not two.
 Sibling discipline: [[rank-on-fields-you-actually-write]] — verify tool field constraints before composing for them, don't re-derive at call-time.
@@ -0,0 +1,23 @@
 ---
 name: verify-counts-two-ways-mechanical-sweeps-are-finicky-in-this-shell
 description: "grep tallies / frontmatter scans misfire in this zsh (glob quirks; all-tree greps catch HTML class= + code-block noise). Verify every count two ways before it enters a record; and don't dismiss correct data on a prior expectation. Three near-false-findings caught this way 2026-05-29."
 metadata: 
  node_type: memory
  type: feedback
  originSessionId: a8eb1d46-314e-4545-a133-f747c69ad5b4
 ---
 Mechanical corpus sweeps (class-tallies, frontmatter scans, "does X exist" greps) are **unreliable in this zsh shell** and must be cross-checked before any count enters an audit or a record.
 **Why:** observed repeatedly on 2026-05-29 during the code-represents-spec audit —
 - `--include=*.md` unquoted → zsh "no matches found" (glob expansion); the tally silently didn't run.
 - `grep -rhoE` across all of `content/` caught non-frontmatter `class:` (HTML `class=` attributes, code blocks) → spurious "types" (offering, lex, not-found, sequence-index) that looked like clause-1 findings.
 - Two runs of "the same" tally disagreed wildly (124 essay vs 193 glimpse) — one was noise, one was right.
 **How to apply:**
 - Prefer **per-pattern `grep -lE '^class:[[:space:]]*X' files | wc -l`** (count files matching one exact pattern) over `-o`/`-m1`/`sed` pipelines.
 - Quote globs: `--include='*.md'`.
 - Scope explicitly (e.g. `content/posts` + `content/pages`, not all of `content/` with its chamber HTML).
 - **Confirm any surprising count a second way** before reporting it — *both* directions: don't assert a finding from a noisy tally, and don't *dismiss* a correct tally because it violates a prior expectation (the "essay must be dominant" anchor nearly buried the true glimpse-dominant count).
 The cost is real (tool-churn), but it is the "once, correctly" tax. Related: [[feedback-trust-prior-pass-frame]] (verify-before-compose); the logged drift `asserting-fs-state-from-a-misread-listing`.
@@ -0,0 +1,16 @@
 ---
 name: verify-each-post-type-after-changes
 description: "After any ARC change touching shared CSS/templates/composition, verify EVERY content type on BOTH desktop and phone — not just the types you think you touched, not on one viewport."
 metadata: 
  node_type: memory
  type: feedback
  originSessionId: c8a6e615-64ae-41ba-a659-49e91fb5af92
 ---
 Steward's standing instruction (2026-06-01): **"always check each post type after any changes we make."**
 After any change to shared SCSS, templates, or composition logic, spot-check **each ARC content type** (essay · meditation · observation · reverie · dream · fragment · glimpse · photo-essay · gloss · readings · about · heteronym · hommage · tombeau, plus Annals view + Chamber seal) on **both desktop and phone viewports** before considering it verified.
 **Why:** The §II.d marginalia-gate (`913aaea`) was verified on *phone* for the types looked at, and only on phone. Its **desktop** rendering for *no-marginalia* text-primary pieces (essay/gloss/readings → body centred at 38rem, a narrow column floating in a wide viewport) went unseen until two sessions later, where it read as "something broke." It hadn't — the gate CSS was byte-identical to the verified state (`git diff 913aaea HEAD -- _sidenotes.scss` empty) — but a per-type × per-viewport pass would have surfaced the desktop composition at authoring time, when the decision was live, instead of as a late surprise. Composition changes ripple by *content type* (the `:has()` gate fires differently per type's marginalia) and by *viewport* (38rem reads as full-width on phone, narrow-centred on desktop) — so a single-type, single-viewport check systematically misses the blast radius.
 **How to apply:** Treat "verified" as covering the full type matrix × {desktop, phone}, not a sample. When a change is theme/composition/cascade-level, enumerate the types (the content-typology list) and the two viewports as an explicit checklist; render-and-look at one representative page per type in each. This is the per-type/per-viewport extension of [[feedback-trust-prior-pass-frame]] (re-run verification at the scope of the change, not the scope of the prior pass) and belongs in the SCSS-refactor render-equivalence verification ladder (skill-harvest candidate). Sibling drift: verifying on the viewport/sample that's convenient, not the one where the change is most visible.
@@ -0,0 +1,16 @@
 ---
 name: arc-content-chamber-is-intentional-v1-chamber-legacy-deferred-presentable-cluster-to-do
 description: "The 39 no-class files under ARC's content/chamber/** are v1-Chamber OUTPUT (fictional canon, deliberations, protocols) — intentional retained legacy, NOT drift. Deferred ARC to-do: gather them into a presentable cluster as a record of the development that informed the still-unbuilt source-aware Chamber. Out of content-typology §5 clause-1 scope. Do not delete, do not reclassify."
 metadata: 
  node_type: memory
  type: project
  originSessionId: a8eb1d46-314e-4545-a133-f747c69ad5b4
 ---
 ARC's `content/chamber/**` (canon/chamber-generated, deliberations, protocols, hermetic, inventions, hybrid, synthesis, contested) is **v1-Chamber output** — the fictional-canon era. It is **intentional retained legacy, not drift**: do not delete, do not force a `class:`, do not treat the 39 missing-`class:` files as findings.
 **The deferred to-do (steward, confirmed 2026-05-29):** gather this original chamber material into a **cluster of pages** that shows what v1 was — a *record of the ongoing development* that informed the design of the **source-aware Chamber** (still unbuilt; depends on the studium-engine maturing). Discussed and deliberately deferred; sits on the long ARC to-do list. Execution not scheduled.
 **Scope ruling for audits:** out of `content-typology-specification.md` §5 clause-1 scope (clause 1 governs §1-typed authored content; chamber files are §2-apparatus / legacy record, no `class:` regime). Recorded in the 2026-05-29 code-represents-spec audit (`docs/AldineXXI-Codex/_audits/code-represents-spec-audit-2026-05-29.md`, Scope section).
 **Distinct from** [[project-chamber-cruft-restoration]] — that is the chamber-**library** (399 source texts: Bachelard, Adorno, …) restoration for "discourse with the library." Different corpus, different work. Don't conflate.
@@ -3742,3 +3742,64 @@ The ELEVATED Stage F item is done and live. Commit **`e1d0f12`** on `feat/compas
 **Returns:** R1 — asserted a font license from memory, not from the file; corrected by reading COPYING (the steward's "how do people use these as emoji?" re-grounded the whole question — symbol/design free, only font software licensed). R2 — honest restraint held (drawn hand named as the weak mark, not oversold). **Discipline:** render-and-look (Chrome-headless → Read PNG, on my own SVGs and the candidates) + measure-the-real-font (Apple Symbols proportions, Garamond coverage) instead of guessing.
 **Follow-on (minor, deferred):** `aria-label`-per-link a11y on the compass (links now rest on `title`; SVG is `aria-hidden`) — on the spec accessibility audit list, not blocking. Reusable method recorded: fontTools extraction from EB Garamond for any typographic sigil; fontforge for Type1 (URW base-35).
 ---
 ## 2026-05-28 (evening) → 2026-05-29 — vignette spec landed (jurist pass) + code-represents-spec audit + Hakyll north-star
 *(Tracker was current through 2026-05-26; this catches up the 2026-05-28 evening + 2026-05-29 arc. Per-session detail in the session memory files.)*
 **2026-05-28 (evening)** — content-types.yml generator landed (`a9dc188`); vignette spec revision brief drafted (`438ee03`, Shape C: Claude Code as Phase-α engine, /vignette skill as codified discipline, interface contract as α→ω bridge). Detail: `session-2026-05-28-evening-arc-stage-f-vignette-scoping.md`.
 **2026-05-29** — Vignette spec open questions settled (`drafts/vignette-decisions-settled-2026-05-29.md`, `d2ca4b0`): dark mode **mode-aware / genome-mode-independent** · reduced-motion **breath-stops** · cycle-end **hold-at-endpoint** · keep **Phase α/ω**. **Jurist revision landed canonical** at `vignette-specification.md` (`c52fb69`) — sharpened the dwell-test (first prototype = mandatory Claude.app pass, against *unfelt* error), the transition trigger (+ L1-stability precondition), the cycle-end clamp (enforces the luminance palette bound), and surfaced the Phase-1 **orientation-preserving luminance** constraint (a dark-mode bug invisible to visual review — must be tested against semantic orientation, not "does it look right").
 Steward reframed sequencing: **audits before the vignette build** (completion + Alexandrian idiom + do-once; Phase 1 is audit-insulated but the deeper reasons hold). **code-represents-spec audit** run (spine = content-typology §5 Code Binding Contract; `_audits/code-represents-spec-audit-2026-05-29.md`, `b15684a`) — §5 largely honored; **two gaps**: clause-2 gloss listing glyph drift FIXED (`164d8c3` ✎→¶; root cause "declared twice" → Hakyll audit), clause-3 silent wildcard fallback → **ships-with-Stage-G** (the spec's own deferred enforcement). Clauses 1/4/5/6/7 pass; 8/9 forward-looking (eligibility machine-verified vs §2). **No spec-gaps → Stage G promotion unblocked.** `content/chamber/**` recorded as intentional v1-legacy (the deferred presentable-cluster to-do), out of clause-1 scope.
 **Hakyll north-star** written (`drafts/the-hakyll-north-star-2026-05-29.md`, `d714a11`) — sibling to scss-cascade-north-star, child of the-tree. The build **reads** content-types.yml as single source rather than re-encoding it: add a type = add a YAML row. Centerpiece = collapse the ~12 near-identical listing `create` blocks into one generator over the table; single-source class→glyph; case-statements → table lookups (unknown = build error). This is where code-represents-spec **clauses 2 & 3 get their durable, structural cure**. CLAUDE.md Hakyll version corrected `4.16.2.1` → `4.16.6.0` (`7bc2a8a`; resolver-pinned note added).
 **State:** ARC `main` @ `7bc2a8a`, **8 ahead of origin** (local-only by directive). **Stage F remaining:** the two Alexandrian audits (Hakyll-simplicity + SCSS-cascade, both north-stars ready) → then the plan to complete the composition work + write/test the vignette system (Phase 1 insulated, post-audits) → **Stage G** (jurist → content-typology OPERATIVE, fed by the code-represents-spec gap-map) closes. **Pulling thread: run the two audits tomorrow.**
 ---
 ## 2026-05-30 — Both Stage-F audits run; three reconciles; the Stage-F-closing plan (vignette → F+1)
 **Both Alexandrian audits written** — gap-maps (verified/inferred per finding; cleanup/table-truthful-first; execution *staged*, not paper-committed). Commit `d230ab1`.
 - **SCSS-cascade** (`_audits/scss-cascade-audit-2026-05-30.md`): diagnosis = partials cut by *page-area*, the dream wants *scale* → `_utilities` dissolves, `_post`/`_base` split, `_layout`/`_navigation` reclaim, `_syntax` rises, `_dark` mostly-vestigial, `_chamber` quarantines (block 2). **Steward decoupled lean vs clear:** cleanup makes it *lean*; re-partition makes it *clear* (line-neutral, source-order risk). Execution = cleanup-first (committed) → **re-read gate** → re-partition (empirical, conditional).
 - **Hakyll-simplicity** (`_audits/hakyll-simplicity-audit-2026-05-30.md`): **FEASIBILITY GATE CONFIRMED** against unpacked 4.16.6.0 source (`create`/`match :: … -> Rules () -> Rules ()`; `Rules` is a monad; `preprocess :: IO a -> Rules a`; codebase already parses YAML). 11 near-identical listing blocks → one generator over content-types.yml. Found a dead disabled-glyph system still in `postCtx`, and table-vs-build drift (§4D).
 **Three reconciles (drift the audits surfaced):**
 - **Compass surfacing** (`ab89d65`): §2 marked dream/heteronym/hommage/reverie `governed-but-unsurfaced` while the build (live `create` blocks) + corpus (1 instance each) had them surfaced. Steward confirmed + ruled **reverie = surfaced**. Fixed §2 → `surfaced` (tombeau stays — no instance); regenerated YAML; `content-types-check` green; diff scoped to the 4 fields. §VI/cross-spec count (now 13 + Annals) → Stage G.
 - **`titleClasses` → user-visible `[FIX]`** (`dea751a`): `showTitle` (post.html `<h1>`) gated on `{essay,meditation,chamber}` while 10 types are `title_display: shown` → gloss/readings/dream/heteronym/hommage/reverie rendered **title-less, against spec**. Expanded to the shown-set (stopgap; durable = table-driven lookup in the generator). Verified via `rebuild-local` (six regained `<h1 post-title>`; not-shown types unchanged). Third clause-2-family instance.
 - **ADR-006 stage tracker** (`990c67d`) — see the rule below.
 **⚠ THE STAGE-E FIASCO → STANDING ANTI-DRIFT RULE.** Steward caught me repeatedly invoking "Stage E discipline." Verified: **Stage E was closed + MERGED** (`c480b09` "peer-spec propagation" → `2a2c195` "hardening" → `558f9f3` "post-Stage-E live state") — but **ADR-006's stage tracker was frozen at authoring-time** (only Stage A ✓ DONE; closing line still "Stage B begins next session"). A *record* gap, not a work gap; my hedge had anchored on the stale 2026-05-19 audit-delta snapshot. Reconciled by adding a dated stage-status block to **ADR-006 Part 5**. **RULE (now standing): ADR-006 Part 5 is the *canonical* stage tracker; this file mirrors it; when a stage closes, update BOTH in the same move.** Also corrected the mislabel — the "regenerate spec+YAML together" rule is the *matrix-as-canonical / YAML-as-shadow invariant* (ADR Part 4, permanent), **not** Stage E.
 **Stage-F-closing plan** (`drafts/stage-f-closing-plan-2026-05-30.md`). Two steward decisions: **vignette Phase 1 → F+1** (not part of closing F); **§II.d marginalia render = SCSS not site.hs** (rides `:has()`, already in use → folds into the SCSS track). Collapses to **two tracks + quick wins**: Phase 0 (titleClasses [FIX] ✓; SCSS RETIRE) · Phase 1 SCSS cascade (incl. the `:has()` marginalia-composition gate) · Phase 2 Hakyll build (table-driven lookups; retire dead glyph; the generator) → **F closes** → Stage G (jurist) → F+1 (vignette). **Marginalia scoped:** current grid is unconditional (empty asymmetric = the "unearned gloss-form"); fix = gate on `:has(marginalia)`, symmetric by default; the §II.d double-center concern is spec-deferred (out of scope).
 **CANONICAL STAGE STATUS (mirror of ADR-006 Part 5, as of 2026-05-31):** **A–E DONE + merged · F in final items** — Phase-1 SCSS cleanup DONE (2026-05-31); remaining = Phase-2 Hakyll generator **· M (mode modernization) pulled forward from F+1** (between F and the full re-partition) **· full re-partition** now follows M (theme-free ground) **· G pending** (jurist promotion; carries the §VI count). Macro-sequence: F Phase-2 → M → re-partition → G → F+1 vignette. *Keep this line in step with ADR-006 Part 5 — the anti-drift rule above.*
 **State (midday):** ARC `main` local-only (was 8 ahead; +`d230ab1` audits · `ab89d65` surfacing · `990c67d` ADR reconcile · `dea751a` titleClasses fix · `90802a6` plan).
 **Afternoon (same session):** Phase 0 **RETIRE** (`5f4c27e` — ~190 dead SCSS lines: voice colours/$enable/metadata/15 mixins; **byte-identical CSS**, zero render) → Phase 1 **DEDUPE** (`dfff604` blockquote/enfilade/small-caps, per-selector-diff verified) + **small-caps drift fix** (`192423e` — one small-caps, footer font-variant→SC face) + **token rename** (`9156522` — `$sidenote-size`→`$marginalia-register-size` to `_variables`). Then **MODE-MECHANISM MODERNIZATION DECIDED** (`4e4f700`; `drafts/mode-mechanism-modernization-2026-05-30.md`) — the thing the steward had circled: **drop the toggle, adopt the system** via `color-scheme`+`light-dark()` (pure CSS, zero JS, no FOUC), **+P3 wide-gamut +`prefers-contrast`**; principle = *honor the reader's environment, never ask them to re-declare it* (the font-size-percentage posture generalized); **founds the vignette (F+1)**; **SHELVES the dark→var MIGRATE** (superseded — light-dark deletes `_dark`+scattered blocks wholesale). **10 commits; ARC `main` 19 ahead, local-only, clean; nothing deployed.** **Revised pulling thread: Phase 1 FIX** (`:has()` marginalia-render gate + §I.h blockquote fix + stale compass `@extend`s + compass a11y) → re-read gate → Stage F closes (all theme-independent; mode rework is F+1).
 ## 2026-05-31 — Phase-1 FIX items landed + mode modernization PULLED FORWARD (Stage M)
 **Session shape:** woke into the confirmed Phase-1-FIX pulling thread; closed all of it; steward then decided to pull the mode modernization forward. **4 commits, ARC `main` ahead 23, local-only, clean, nothing deployed.**
 **Committed (all verified; gate+glimpse+compass-a11y confirmed on the steward's physical phone, a11y via VoiceOver):**
 - `b228f47` — removed stray `_layout.scss.bak` (240 dead lines, 11 months old in live tree).
 - `913aaea` — **§II.d `:has()` marginalia-render gate** (the substantial Phase-1 FIX). Grid renders only when a piece carries marginalia; no-marginalia pieces compose **symmetric, body centred at 38rem** (spec L260 read verbatim — NOT 720px; `.page-content`'s 720 was the wrong fallback). Symmetric base + `:has()`-gated grid (wins by specificity) + glimpse guard (`max-width:none`). Verified desktop + phone.
 - `f390b5d` — **glimpse phone full-bleed fix** (pre-existing LIVE bug; steward confirmed on phone). Root cause = TWO clips on the `100vw`+negative-margin burst at ≤767px: the parent `<figure>` (`overflow-x:auto` → clip+scroll) AND `.wrapper/.page-content/main` (`overflow-x:hidden` → clip). **Both must open together** (figure-only=padded; wrapper-only=padded+scroll), scoped `body.glimpse-page` + `.glimpse figure`. Took 3 attempts — two wrong diagnoses first (reasoned from a comment; then asserted the wrapper was the sole clip) before reading the full ancestor chain. Lesson reinforced thrice this session: **read rendered output / full DOM chain before asserting a CSS causal mechanism.**
 - `02393bc` — **compass §3F + §3G** (one commit; two concerns entangled in `_compass.scss`). §3F: removed the dead `.path-*` `@extend .current` block (all 7 inert — Jekyll-era mechanism; body class is now `<class>-page`; `/archive/` `/chamber/` hrefs themselves stale; verified via git pickaxe + built HTML). §3G: compass a11y — visually-hidden real text node (`.compass-label` via the preserved `@mixin visually-hidden`) per point. **Chose visually-hidden text over `aria-label`** for durability: translatable (ARC multilingual), survives CSS-off, native-semantic.
 **§I.h blockquote:** deferred to ride Stage M (steward 2026-05-31, Option B) — its clean form is the consolidated single source after `_dark.scss` is gone; doing it now = a `_dark` edit deleted weeks later (anti-do-once).
 **⚠ THE BIG DECISION — MODE MODERNIZATION PULLED FORWARD (Stage M).** Steward chose to do the mode work **now**, not at the start of F+1. This **resolves a contradiction the closing plan carried**: §5b said modernization should precede the re-partition (so it doesn't arrange soon-deleted SCSS), but §6 listed the full re-partition as a *Stage-F* closer while modernization sat in *F+1* — the re-partition couldn't both close F and follow an F+1 step. Pulling modernization forward (Stage M, between F and the re-partition) fixes it: theme layer deleted first, re-partition + F+1 vignette built on theme-free ground **once** (do-once). **Full re-partition is no longer a Stage-F deliverable** — it follows M. **Plan amended this session** across all three records (anti-drift rule): `stage-f-closing-plan-2026-05-30.md` (new §5c + §6 revised), ADR-006 Part 5 F-line, this tracker. Scope surveyed (bounded): ~12 tokens → `light-dark()`; `_dark.scss` (146 lines) deleted; exactly 7 scattered `[data-theme="dark"]` blocks folded in; toggle+JS removed. Staged M(a) structural core → M(b) P3 (steward has M1 Air + iPhone, both P3) → M(c) `prefers-contrast`.
 **Process note (contamination/flakiness):** the tool-output channel flaked intermittently this session; early on I **confabulated unread spec + SCSS content over empty tool returns twice** (caught + discarded + re-read from verified output). Plus the glimpse double-mis-diagnosis. Through-line discipline: read the actual rendered/built artifact; an empty tool return is not a read; verify a CSS causal story against the DOM before asserting it.
 **Pulling thread → Stage M(a) fresh next session** (the structural core: `light-dark()` migration + delete `_dark`/7 blocks/toggle; both-mode verify by OS-appearance toggle on real devices). Then re-partition, then G, then F+1 vignette.
 **Literal question for next session:** start M(a) by migrating the `main.scss` token block to `light-dark()` first (then delete `_dark` + the 7 scattered blocks + toggle, verify both modes), or inventory all 7 scattered blocks' non-flip dark choices *first* so none is lost in the fold? And: M(a) as one commit or split (token-migration / delete-dark-layer / remove-toggle-JS)?
 **Standing flag:** the glimpse fix corrects a **live reader-facing bug** still on the deployed site (every glimpse, every phone) — committed but not deployed. Whenever the steward next deploys, that one has real urgency. Deploy remains the steward's deliberate act; ARC `main` is 23 ahead, local-only.
@@ -0,0 +1,98 @@
 ---
 name: session-2026-05-29-arc-vignette-spec-landed-jurist-pass-code-represents-spec-audit-hakyll-north-star
 description: "Woke into the vignette Phase 1 thread; instead the whole session became the Stage-F-closing groundwork. Settled the vignette spec's open questions → jurist revision landed canonical (c52fb69). Steward reframed sequencing: audits before the vignette build. Ran the code-represents-spec audit (§5 spine; b15684a) — §5 largely honored, two bounded gaps (clause-2 glyph drift FIXED 164d8c3; clause-3 silent fallback ships-with-Stage-G). Wrote the Hakyll north-star (d714a11). Pulling thread: run the Hakyll-simplicity + SCSS-cascade audits tomorrow — they are the gate to the plan that completes composition, writes+tests the vignette system, and closes Stage F."
 metadata: 
  node_type: memory
  type: project
  originSessionId: a8eb1d46-314e-4545-a133-f747c69ad5b4
 ---
 # Session 2026-05-29 — ARC: vignette spec landed + code-represents-spec audit + Hakyll north-star
 **Wake context.** /wake-up + /remote-control into the vignette Phase 1 thread (16h pause). Symmetria init closed the wake's clasp. MemPalace healthy (45,898 drawers). The session did NOT go to Phase 1 — it became the Stage-F-closing groundwork instead, by the steward's sequencing reframe.
 ## Past — what we did (chronological)
 ### 1. Vignette spec: alignment → open questions settled → jurist revision landed
 - **Alignment exercise** (steward asked me to explain what vignettes are before building) — confirmed shared understanding against the actual spec (read fresh, not from memory): threshold object before the title; four layers (ground mark / geometry / field / breath); the interface contract as the Phase-α→ω bridge.
 - **Settled the brief's 7 open questions** with the steward. Firm: dark mode → **field mode-aware, genome stays mode-independent** (rationale: mode is a property of the *reading* not the text; don't bloat the contract — the renderer owns presentation; ground mark already mode-relative by construction); reduced-motion → **breath stops, static frame**; cycle-end → **hold at trajectory endpoint**; naming → **keep Phase α/ω**. Wrote `vignette-decisions-settled-2026-05-29.md` (**d2ca4b0**).
 - Assembled the jurist packet (3 essential + supporting + reference tiers; verified all paths).
 - **Jurist returned the revised spec → landed canonical** at `docs/AldineXXI-Codex/vignette-specification.md` (**c52fb69**) + CLAUDE.md companion-pointer updated. The jurist went *past* our dispositions in three places, two of which corrected my leans:
  - **Dwell-test interface:** I'd leaned hybrid (escalate on felt uncertainty); jurist made the **first prototype a mandatory Claude.app pass regardless of confidence** — "an escalation rule keyed only to *felt* uncertainty is structurally blind to *unfelt* error." I'd have shipped the weaker version.
  - **Transition trigger:** added an objective **L1-stability precondition** under my qualitative gate.
  - **Cycle-end** is now load-bearing for the luminance palette bound (a render-time **clamp**, mandatory not aesthetic).
  - **The Phase-1 trap (most important for the future build):** the renderer's luminance mapping must be **orientation-preserving, NOT a naive offset from current ground luminance** — a naive offset inverts the luminance axis in dark mode and runs a text's temporal direction backwards, *and the dwell-test wouldn't catch it because the result still looks plausible*. A bug that hides from visual review → must be tested against *semantic* orientation, not "does it look right."
 ### 2. Sequencing reframe (steward): audits before the vignette build
 - I had *started* Phase 1 (read-only recon only — survey + tooling; zero code). Steward: "is this the right order? we have major audits to finish Stage F." Weighed honestly — Phase 1 is *audit-insulated* (isolated renderer in `dev/`, reads tokens at runtime, ends at dwell-test gate, touches neither cascade nor site.hs) so there's no *technical* collision — but **agreed**: completion + Alexandrian idiom + do-once favor audits-first. Reordered at zero sunk cost. (Recon preserved: exact palette tokens, `[data-theme="dark"]` mechanism, `dev/` safe-harbor, Chrome-headless render path — for the eventual build.)
 ### 3. code-represents-spec audit (spine = content-typology §5 Code Binding Contract)
 - Steward chose this audit first ("code should match spec before making it Alexandrian"; truth before arrangement). Scope = §5 as spine; run against §5-as-authored to **feed Stage G's jurist promotion**. Symmetria check walked → proceed; deliverable reframed as a **gap-map** (not pass/fail).
 - All 9 clauses traced. **Result: §5 largely honored — TWO gaps only:**
  - **Clause 2** — gloss listing-header glyph still ✎ (`site.hs:899`) while Compass migrated to ¶ (`e1d0f12`); class→glyph declared twice and drifted. → **[FIX] done: 164d8c3** (✎→¶). Root cause (declared twice) → Hakyll audit single-sourcing.
  - **Clause 3** — `classRouteName`/`classDisplayName` end in silent wildcard fallback (`other -> other ++ "s"`) instead of build-error. → **ships-with-Stage-G** (the spec's own deferred enforcement item 9; low-risk, explicit cases already total).
  - Clauses 1/4/5/6/7 PASS; 8/9 forward-looking (eligibility data machine-verified vs §2 via `make content-types-check`). **No spec-gaps → Stage G promotion unblocked** once the two items are scheduled.
 - Deliverable: `docs/AldineXXI-Codex/_audits/code-represents-spec-audit-2026-05-29.md` (**b15684a**).
 - **Chamber scope:** confirmed `content/chamber/**` (39 no-class files) = intentional v1-Chamber legacy (the deferred presentable-cluster to-do), **out of clause-1 scope** — not findings. Captured `project-arc-chamber-v1-legacy-cluster.md` + MEMORY pointer (distinct from chamber-*library* restoration).
 ### 4. Hakyll north-star (the build-layer orientation)
 - Steward confirmed the Hakyll-simplicity audit's three framings: same reach as SCSS (deep look, ideal expression, "the tree not its atoms"); content-type addition = **a new YAML row** ("spec so clear code naturally flows from it"); build layer only.
 - Wrote `docs/AldineXXI-Codex/drafts/the-hakyll-north-star-2026-05-29.md` (**d714a11**) — sibling to scss-cascade-north-star, child of the-tree. Core claim: **the build *reads* the type system (content-types.yml) as single source rather than re-encoding it.** Centerpiece: collapse the ~12 near-identical listing `create` blocks into one generator over content-types.yml; single-source class→glyph; case-statements → table lookups (unknown = build error). This is where code-represents-spec **clauses 2 & 3 get their durable, structural cure** (not patches — a structure that can't drift).
 - **Hakyll source-learning folded into the audit run** (not pre-staged): `stack unpack hakyll-4.16.6.0` at audit step 0, read `Hakyll.Core.Rules` + `Context` to confirm the data-driven Rules affordance.
 ### 5. CLAUDE.md Hakyll version fix
 - Discovered the pinned version is **4.16.6.0** (resolver lts-23/24, GHC 9.8.4), not the **4.16.2.1** CLAUDE.md claimed "at last check." Fixed (**7bc2a8a**) + added a resolver-pinned note + verify command so the number can't masquerade as authoritative and go stale again. Assessed: **no Hakyll/resolver upgrade warranted** — ARC uses the long-stable core; the audit is version-agnostic; the version is resolver-pinned (bumping = bumping the whole snapshot); even the SCSS-partial-mtime friction is a design characteristic, not version-fixable.
 ## Present — mood / returns (from ledger)
 - **The session's hallmark was verify-don't-assert.** Three near-misses, all caught before entering the record: (1) nearly reported false clause-1 findings (chamber/reading/photo) from a noisy all-`content/` grep tally; (2) then nearly *dismissed the correct* clean tally (193 glimpse / 9 essay) on a false prior that essay is corpus-dominant; (3) floated a false clause-7 "no redirect" gap (static/_redirects exists with proper 301s). Each resolved by a controlled second check.
 - **Challenge-the-framing operative:** on dark mode I did not rubber-stamp the steward's guess of option (c); reasoned to (a) on contract-bloat + semantics/presentation grounds; steward agreed.
 - **Named honestly that the jurist sharpened my leans** (dwell-test, transition trigger) — I'd have shipped weaker versions. The loop is load-bearing; the jurist caught what I couldn't from where I sat.
 - **Load-bearing:** the audit's gap-map shape (two bounded gaps, classified, feeding Stage G) and the Hakyll north-star's central claim (build reads spec, can't drift). The convergence §5 → content-types.yml → data-driven build is the through-line of the whole session.
 ## Confidence to recalibrate (for next session)
 - **Mechanical bash sweeps are finicky in this shell** (zsh glob quirks; all-tree greps catch HTML `class=` + code-block noise). Verify every count two ways before it enters a record. Prefer per-pattern `grep -lE` file counts over `-o`/`-m1` pipelines.
 - **Anchoring on a prior expectation can make you dismiss correct data** (the essay-dominant assumption). Verify both directions.
 - **"It's in memory somewhere" may be adjacent-but-distinct** — chamber-*library* restoration ≠ ARC-content-chamber legacy cluster. Distinguish; don't conflate.
 ## Future — what is pulling
 ### Pulling thread (singular)
 **Run the two Alexandrian audits tomorrow — Hakyll-simplicity (its north-star is written and ready) and SCSS-cascade (north-star ready) — because together they are the gate to the plan that completes the composition work, writes+tests the vignette system, and finally closes Stage F.** (Steward's own framing this wrap.) The audits are not a detour from Stage F; they are how Stage F closes cleanly, on audited ground, in the settled Alexandrian idiom.
 ### Actionable resumption point (as of wrap — re-judge against what changed)
 - ARC `main` @ **7bc2a8a**, **8 ahead of origin** (local-only by directive). **Working tree clean.**
 - Read-first artifacts: `the-hakyll-north-star-2026-05-29.md`, `the-tree-north-star-2026-05-25.md`, the code-represents-spec audit deliverable (its clause-2/3 findings get their durable cure in the Hakyll audit), and `scss-cascade-north-star-2026-05-19.md`.
 - **Candidate first move:** begin the **Hakyll-simplicity audit** per the north-star's §Sequencing. Step 0 = `stack unpack hakyll-4.16.6.0` to a scratch dir **outside the ARC tree** (so it's never built); read `Hakyll.Core.Rules` + `Hakyll.Web.Template.Context` to **confirm the data-driven Rules affordance** (the one feasibility gate). Then read `site.hs` top-to-bottom, mapping every per-type-fact and per-type-repeated-block; design the single generator over content-types.yml.
 - **Alternative first move:** run the **SCSS-cascade audit** first (sibling; no dependency; north-star already written). Steward's pick — they don't collide.
 ### Literal question for next-Claude
 **Which audit does the steward want to lead tomorrow — Hakyll-simplicity or SCSS-cascade?** And the real feasibility gate beneath the Hakyll one: **does Hakyll 4.16.6.0's `Rules` monad actually confirm the data-driven generation the north-star bets on** (generate the listing rules by iterating a parsed content-types.yml, YAML parsed in `main`)? Until that's verified against the unpacked source, the audit's centerpiece is a high-confidence assumption, not a confirmed affordance — confirm it at step 0 before designing on it.
 ### Pause statement
 I'm about to be away. I don't know whether the steward will lead with Hakyll or SCSS tomorrow, or whether anything shifts overnight. What I want to find still pulling: the two audits as the gate to the Stage-F-closing plan — run on audited ground, in the Alexandrian idiom, with the code-represents-spec gap-map (clauses 2/3) getting their durable cure in the Hakyll audit. The link is actionable: the north-star is written, step 0 is named, the tree is clean.
 ### Decisions deferred (and why)
 - **The two audit RUNS** — tomorrow (steward's call: "run the audits tomorrow").
 - **The clause-3 build-error conversion** — bundle with the Stage G §5→OPERATIVE commit (the contract's own sequencing), not before.
 - **Stage G jurist promotion** of content-typology → OPERATIVE — the audit feeds it; jurist territory.
 - **The vignette Phase 1 build** — after the audits, per the agreed sequence (Phase 1 is insulated but completion-first holds).
 - **Push ARC to origin** — local-only by directive; 8 ahead; push at a natural moment.
 - **The chamber v1-cluster presentable pages** — long-deferred ARC to-do; not this arc.
 - **No Hakyll/resolver upgrade** — no ARC pain it fixes; audit is version-agnostic.
 ## Skill harvest (§1.6) — proposals for steward authorization (never autonomous)
 **New candidate this session:**
 - **`/spec-code-audit` (or `/contract-audit`)** — the code-represents-spec method generalized: audit live code against a spec-contract section *clause-by-clause*; classify each gap (FIX-now / route-to-authority / ships-with-promotion); verified-vs-inferred marked per finding; deliverable as a **gap-map**, not pass/fail. Hard-won this session and immediately reusable — the SCSS-cascade and Hakyll audits share the gap-map + verify-don't-assert discipline, and it applies beyond ARC (L1 code-vs-amendment). *Propose for authorization.*
 **Carried forward, still unauthorized** (from prior wraps): `bmf-diagnose`, `/arc-typeset`, `/pre-build-audit`, the Symmetria §3 `trust-prior-pass-frame` flag, and `/vignette` (Phase 2 of the vignette build, post-audits).
 **Saved as feedback memory (not a skill — within discipline):** the verify-counts-two-ways / mechanical-sweeps-finicky-in-zsh lesson.
 ## Pointers
 - 6 commits this session on `animal-davidglidden-eu/main` (local-only): d2ca4b0, c52fb69, b15684a, 164d8c3, d714a11, 7bc2a8a. Plus the prior session's 438ee03 + a9dc188 = 8 ahead of origin.
 - Vignette spec now canonical (revised); brief + decisions-note preserved as bridge artifacts.
 - ARC tracker `project-arc-rework.md` appended this wrap (was behind through 2026-05-26).
 - Session ledger: `session-ledger-2026-05-29.md`.
@@ -0,0 +1,84 @@
 ---
 name: session-2026-05-30-arc-stage-f-both-audits-run-three-reconciles-retire-dedupe-mode-mechanism-modernization-decided
 description: "Woke into the two-audit Stage-F gate; ran both (Hakyll feasibility CONFIRMED). Three drifts reconciled — Compass surfacing, the ADR Stage-E record fiasco (+ standing anti-drift rule), and a user-visible titleClasses bug. Drew the Stage-F-closing plan (two tracks; vignette→F+1; marginalia=SCSS via :has()). Executed Phase 0 RETIRE (~190 dead lines, zero render) + Phase 1 DEDUPE (blockquote/enfilade/small-caps) + small-caps drift fix + token rename. Then the steward's long-standing question resolved: modernize the dark/light mechanism — drop the toggle, adopt the system (color-scheme + light-dark) + P3 + prefers-contrast, under 'honor the reader's environment'; founds the vignette; SHELVES the dark→var MIGRATE. Pulling thread: Phase 1 FIX (the :has() marginalia gate + §I.h blockquote + compass cleanup) → re-read gate → Stage F closes."
 metadata: 
  node_type: memory
  type: project
  originSessionId: 13e0d4eb-b3bf-42ef-ac94-d642eff3ec3d
 ---
 # Session 2026-05-30 — ARC Stage F: audits, reconciles, RETIRE/DEDUPE, mode-mechanism modernization
 **Wake context.** /wake-up + /remote-control into the confirmed pulling thread: run the two Alexandrian audits (Hakyll-simplicity + SCSS-cascade) = the gate to closing Stage F. Symmetria active. ARC `main` @ `7bc2a8a`, 8 ahead, local-only. The session went far past the audits — into reconciles, two execution passes, and a foundational architecture decision. **10 commits; main now 19 ahead, local-only, nothing deployed. Working tree clean.**
 ## Past — what we did (chronological, with commits)
 1. **Both Alexandrian audits run + written** (`d230ab1`, gap-maps):
   - **SCSS-cascade** (`_audits/scss-cascade-audit-2026-05-30.md`): diagnosis = partials cut by *page-area*, the dream wants *scale* → `_utilities` dissolves, `_post`/`_base` split, `_layout`/`_navigation` reclaim, `_syntax` rises, `_dark` vestigial, `_chamber` quarantines (block 2). **Steward decoupled lean vs clear:** cleanup makes it lean; re-partition makes it clear (line-neutral, source-order risk). Execution = cleanup-first (committed) → **re-read gate** → re-partition (empirical/conditional).
   - **Hakyll-simplicity** (`_audits/hakyll-simplicity-audit-2026-05-30.md`): **FEASIBILITY GATE CONFIRMED** against unpacked Hakyll 4.16.6.0 source (`create`/`match :: …→Rules ()`; `Rules` is a monad; `preprocess :: IO a→Rules a`; codebase already parses YAML). 11 listing blocks → one generator over content-types.yml. Found: dead disabled-glyph system in `postCtx`; table-vs-build drift (§4D).
 2. **Three reconciles (drift the audits surfaced):**
   - **Compass surfacing** (`ab89d65`): §2 marked dream/heteronym/hommage/reverie `governed-but-unsurfaced` while the build (live `create` blocks) + corpus (1 instance each) had them surfaced. Steward confirmed + ruled **reverie = surfaced**. Fixed §2 → `surfaced` (tombeau stays — no instance); regenerated YAML (`make content-types`); `content-types-check` green; diff scoped to 4 fields. §VI count (now 13 + Annals) → Stage G.
   - **ADR-006 stage tracker / THE STAGE-E FIASCO** (`990c67d`): Steward caught my repeated "Stage E discipline." Verified: **E was closed + MERGED** (`c480b09`→`2a2c195`→`558f9f3`) but ADR-006 Part 5 was frozen at authoring-time (only Stage A ✓). A *record* gap, not a work gap — my hedge had anchored on the stale 2026-05-19 audit-delta. Added a dated stage-status block (A–E done+merged, F in-final, G pending). **Standing anti-drift rule: ADR-006 Part 5 is canonical; `project-arc-rework.md` mirrors it; update both at stage close.** Corrected the mislabel (regenerate-together = the matrix/shadow *invariant*, not Stage E).
   - **titleClasses user-visible bug** (`dea751a`): `showTitle` (the `<h1 post-title>`) gated on `{essay,meditation,chamber}` while §2 marks 10 types `title_display: shown` → gloss/readings/dream/heteronym/hommage/reverie rendered **title-less, against spec** (verified against built HTML). Expanded `titleClasses` to the shown-set (stopgap; durable = table-driven in the generator). Third clause-2-family instance.
 3. **Stage-F-closing plan** (`90802a6`, `drafts/stage-f-closing-plan-2026-05-30.md`): two tracks (SCSS + Hakyll) + quick wins; **vignette Phase 1 → F+1**; **§II.d marginalia render = SCSS not site.hs** (rides `:has()`). Marginalia scoped: current grid is unconditional (empty asymmetric = "unearned gloss-form"); fix = gate on `:has(marginalia)`, symmetric by default; §II.d double-center concern spec-deferred.
 4. **Phase 0 RETIRE** (`5f4c27e`): removed ~190 lines of grep-confirmed dead SCSS — voice colours, `$enable-*` flags, framework metadata (+ the commented debug block), and **15 zero-`@include` mixins** (kept `visually-hidden` as the compass-a11y tool). Gate: rebuild-local compiled + `style.css` **byte-identical** to baseline (zero render change).
 5. **Phase 1 DEDUPE (part)** (`dfff604`): blockquote (removed `_base` dup → `_typography` canonical), enfilade (`.about-enfilade, .chamber-enfilade` merged; chamber's `.back-to-about` kept), small-caps (`.small-caps` class → `@include small-caps`). Verified render-equivalent via per-selector CSS diff (chamber-enfilade preserved 51→51 incl mobile @media; small-caps byte-identical).
 6. **small-caps drift fix** (`192423e`): footer `.site-footer .whisper .small-caps` used `font-variant` (a different mechanism) — steward: "only one type of small caps" → removed; footer falls through to the canonical SC-face rule. The only CSS delta vs baseline.
 7. **token rename** (`9156522`): `$sidenote-size` → `$marginalia-register-size`, relocated to `_variables.scss` (tokens layer); render-identical.
 8. **Mode-mechanism modernization DECIDED + captured** (`4e4f700`, `drafts/mode-mechanism-modernization-2026-05-30.md`): **the thing the steward had been wanting to resolve.** He asked if the dark/light mechanism could be improved; tracing it surfaced 3 seams (auto doesn't follow system on most pages; FOUC risk; two divergent JS paths) + vestigial `_dark.scss`. His **font-size-percentage analogy → the PRINCIPLE: honor the reader's environment; never ask them to re-declare what their system already knows.** DECIDED: **drop the toggle, adopt the system** via `color-scheme: light dark` + `light-dark()` (pure CSS, zero JS, no FOUC). **+ P3 wide-gamut + `prefers-contrast` IN SCOPE** (steward "definitely want p3 ... and contrast"). The web preference matrix (contrast / P3 / inverted-colors / reduced-motion; ambient-light = the unsupported gap) composes with `light-dark()`. **Founds the vignette** (most mode-coupled object; the frontispiece divergence would bite it). **Do-once consequence: the dark→var MIGRATE is SHELVED** (light-dark deletes `_dark` + scattered blocks wholesale — caught before investing). Plan updated §5b.
 9. **Hakyll audit kept truthful** (`<final commit>`): §4D titleClasses confirmed (rendered-output-verified) + surfacing reconcile DONE; §4F mislabel fixed; §6 step-1 surfacing-done.
 ## Present — mood / returns (from the Symmetria ledger)
 - **The session's discipline was verify-don't-assert, ladder-style:** byte-identical CSS gate for RETIRE (dead code → no output change); per-selector CSS diff for DEDUPE (consolidation → computed-equivalent); analytic var-pair reasoning for the migration; rendered-HTML check for the titleClasses bug; unpacked-source verification for the Hakyll feasibility gate. Each verification matched to the kind of change.
 - **My drift, named:** I hedged "can't confirm Stage E closed" by anchoring on the stale 2026-05-19 audit-delta snapshot (and grepping docs not git first). The steward's "we did merge" + git history corrected it. Drift family: *anchoring-on-stale-snapshot* (sibling of trust-prior-pass-frame).
 - **Challenge-the-framing / name-what-you-see operated well:** surfaced the titleClasses bug, the Stage-E record drift, the blockquote-`_dark` override, and — load-bearing — recognized the mode-mechanism question as design-decision-laden (not mechanical) and **surfaced it rather than blasting through**, which produced the do-once catch (shelving dark→var before pouring in throwaway work). The collaboration produced the synthesis: the steward's font-size analogy → the "honor the reader's environment" principle.
 - **Load-bearing return of the day:** *don't do work a known-better approach will redo* — the mode modernization superseding the dark→var MIGRATE was caught at the right moment (during the migration, before most of it).
 ## Confidence to recalibrate (for next session)
 - **DEDUPE/MIGRATE verification is NOT byte-identical like RETIRE** — collapsing/merging changes the CSS *text* even when render is identical. Use per-selector diff (DEDUPE) or analytic value-pair matching (migration), not the hash gate. Remember which gate fits which change.
 - **`_dark.scss` is mostly top-level (NOT `[data-theme=dark]`-scoped)** — so its rules affect *both* modes; several are unintended both-mode overrides (blockquote, kbd, ornament, callout). This is why the mode rework (light-dark) supersedes piecemeal migration.
 - **`make rebuild` / `make build` / `make dev` AUTO-DEPLOY** (CloudFlare + mirror). Use **`make rebuild-local`** (no deploy) for local verification — it exists and is the right path. Confirmed this session.
 - **Exact-match Edits failed twice on blank-line trailing whitespace** — `cat -A` the target region before editing fragile blocks; it paid off.
 ## Future — what is pulling
 ### Pulling thread (singular)
 **Phase 1 FIX — the theme-independent SCSS fixes that carry Stage F toward close:** the `:has()` marginalia-render gate (§II.d composition — gate the grid on marginalia presence; symmetric by default), the **§I.h blockquote fix** (no border, indent both sides, body colour, no forced italic — spec compliance; it also removes the blockquote border drift regardless of the mode rework), the stale compass `@extend`s (`_compass.scss` `.path-chamber`/`.path-archive` → retired routes), and compass a11y (visually-hidden labels — `@mixin visually-hidden` is waiting for exactly this) → then the **re-read gate**. All theme-independent; the mode modernization is shelved to F+1.
 ### Actionable resumption point (as of wrap — re-judge against what changed)
 - ARC `main` @ **`<final>`**, **19 ahead of origin**, local-only by directive, **working tree clean**. Nothing deployed.
 - Read-first: `drafts/stage-f-closing-plan-2026-05-30.md` (§3 marginalia scope; §5b the shelving) + `_audits/scss-cascade-audit-2026-05-30.md` (§3B/E for the FIX items).
 - **Candidate first move:** the **`:has()` marginalia-render gate** in `_sidenotes.scss` (the most substantial FIX, the §II.d composition fix — gate `.bears-apparatus`/`.post-content` grid on `:has(.sidenote, .register-mark)`, symmetric otherwise; `:has()` already used in that partial). Verify across piece types (with-marginalia stays asymmetric; without → symmetric-centered; marker-anchored variant still works; mobile stacks). Render-sensitive → render-and-look, not just hash.
 - **Alternative first move:** the **§I.h blockquote fix** (smallest, clear spec) — but it's *visible* (every quote loses its border + italic), so confirm whether the steward wants to eyeball the render before it commits.
 ### Literal question for next-Claude
 **Which Phase-1 FIX leads — the `:has()` marginalia-render gate or the §I.h blockquote fix?** And for §I.h blockquote specifically: does the steward want to *see it rendered* (it visibly changes every block quote — border + forced-italic both gone) before it's committed, or proceed on the spec's authority? (Both are theme-independent and safe to do before the F+1 mode rework.)
 ### Pause statement
 I'm about to be away. I don't know what shifts overnight. What I want to find still pulling: **Phase 1 FIX as the path to closing Stage F** — on the now-honest table and the cleaned (RETIRE'd, partly-DEDUPE'd) surface — with the **mode-mechanism modernization waiting, fully captured, as the foundation the F+1 vignette arc opens on.** The session resolved a thing the steward had circled for a while; the link is clean — the plan is written, the surface is cleaner, the next FIX items are named and theme-independent.
 ### Decisions deferred (and why)
 - **Mode-mechanism modernization** — decided *in principle* (drop toggle, adopt system, light-dark + P3 + contrast); implementation deferred to **F+1** (the vignette's foundation). Not now.
 - **dark→var MIGRATE — SHELVED** (superseded by the mode rework; light-dark deletes `_dark` + scattered blocks wholesale).
 - **The re-partition** — conditional, at the re-read gate (after the FIX items); the mode rework simplifies it (deletes the theme layer), so cleanest done *before* the re-partition.
 - **gloss-source≡readings-source dedup; compass-point dedup** — deferred to the re-partition (apparatus / nav gather).
 - **§I.h blockquote** — decided to fix (spec compliance); theme-independent, can do in Phase 1 FIX; pending the steward's "see it first?" preference.
 - **Push / deploy** — local-only by directive; 19 ahead; push/deploy at a deliberate moment (his call).
 - **The JS theme-init divergence** (default.html inline auto→light vs theme-toggle.js auto→system) — noted; resolved by the F+1 mode rework.
 - **`_chamber.scss` / chamber v1** — block 2 (separate, content + SCSS together); not Stage-F-blocking.
 ## Skill harvest (§1.6) — proposals for steward authorization (never autonomous)
 **New candidate this session:**
 - **SCSS-refactor render-equivalence verification ladder** — the hard-won discipline of matching the *verification* to the *kind of change*: byte-identical CSS for dead-code removal (RETIRE); per-selector CSS diff for consolidation (DEDUPE); analytic value-pair matching for theme migration; rendered-HTML check for behavior changes. It will recur immediately in the Phase-1 FIX + re-partition work. *Propose: fold into `/spec-code-audit` (or a short `/scss-refactor-verify` note) so it isn't re-derived.*
 **Carried forward, still unauthorized** (from prior wraps): `/spec-code-audit`, `bmf-diagnose`, `/arc-typeset`, `/pre-build-audit`, `/vignette` (Phase 2 of the vignette build), the Symmetria §3 `trust-prior-pass-frame` flag (+ the new `anchoring-on-stale-snapshot` sibling).
 ## Pointers
 - **10 commits this session** (ARC `main`, local-only): `d230ab1` audits · `ab89d65` surfacing · `990c67d` ADR reconcile · `dea751a` titleClasses fix · `90802a6` plan · `5f4c27e` RETIRE · `dfff604` DEDUPE · `192423e` small-caps · `9156522` token · `4e4f700` mode note + plan §5b · + the Hakyll-audit-truthful commit.
 - New artifacts: `drafts/stage-f-closing-plan-2026-05-30.md`, `drafts/mode-mechanism-modernization-2026-05-30.md`, `_audits/scss-cascade-audit-2026-05-30.md`, `_audits/hakyll-simplicity-audit-2026-05-30.md`.
 - ADR-006 Part 5 now carries the canonical stage-status block; `project-arc-rework.md` mirrors it (2026-05-30 section + the anti-drift rule).
 - Session ledger: `session-ledger-2026-05-30.md` (rich — returns, the Stage-E-drift correction, the mode-decision).
 - **MemPalace MCP disconnected mid-session** — drawer/diary/KG filing could not run this wrap; this file + the ledger carry the record. Re-file to MemPalace next session if reconnected.
@@ -0,0 +1,77 @@
 ---
 name: session-2026-05-31-arc-stage-f-phase-1-fix-items-landed-mode-modernization-pulled-forward-stage-m
 description: "Woke into the confirmed Phase-1-FIX pulling thread; closed all of it (marginalia :has() gate, glimpse phone full-bleed, dead compass @extends, compass a11y); §I.h blockquote deferred to ride Stage M. Steward then DECIDED to pull the mode-mechanism modernization forward (Stage M, beside F) — resolving a contradiction the closing plan carried. Plan amended across all three records (anti-drift rule). 5 commits, ARC main ahead 24, local-only, nothing deployed. Pulling thread: Stage M(a) — the light-dark() structural core — fresh next session."
 metadata: 
  node_type: memory
  type: project
  originSessionId: 17ef4cde-4d62-41e8-8098-52898f013b43
 ---
 # Session 2026-05-31 — ARC Phase-1 FIX items + mode modernization pulled forward (Stage M)
 **Wake context.** /wake-up + /remote-control into the confirmed Phase-1-FIX pulling thread (Stage F close). Symmetria active. ARC `main` @ `6279461`, 19 ahead, local-only. The session closed all Phase-1 FIX items, then took a sequencing decision (pull mode modernization forward). **5 commits; main now ahead 24, local-only, working tree clean, nothing deployed.**
 ## Past — what we did (chronological, with commits)
 1. **Spec re-read + §I.h decision (DEFER).** Re-read §I.h Block Quote verbatim (`AldineXXI-specification.md` 771–845) + every real blockquote SCSS rule. Found §I.h is NOT a small fix — it's the SCSS-cascade duplication (5 rules; the `_dark.scss:50` non-mode-scoped block wins both modes per our own `_base.scss:120` note + carries the spec-forbidden `opacity:0.85` luminance attenuation). **Steward chose Option B: defer §I.h to ride Stage M** — its clean form is the consolidated single source after `_dark` is gone; doing it now = a `_dark` edit deleted weeks later (anti-do-once).
 2. **`b228f47`** — removed stray `_layout.scss.bak` (240 dead lines, dated 3 Jul 2025, ~11 months in the live tree; git-tracked → `git rm`; verified unreferenced).
 3. **`913aaea` — §II.d `:has()` marginalia-render gate** (the substantial Phase-1 FIX). Grid renders only when a piece carries marginalia; no-marginalia → **symmetric, body centred at 38rem** (spec L260 read verbatim; the 720px `.page-content` fallback would have been off-spec). Symmetric base rule + `:has()`-gated grid (wins by specificity, `max-width:none`) + glimpse guard. Verified compiled CSS + render on desktop **and physical phone**.
 4. **`f390b5d` — glimpse phone full-bleed fix** (PRE-EXISTING LIVE bug; steward confirmed on phone + screenshot). Root cause = TWO clips on the `100vw`+negative-margin burst at ≤767px: parent `<figure>` (`overflow-x:auto` → clip+scroll) AND `.wrapper/.page-content/main` (`overflow-x:hidden` → clip). **Both must open together** (figure-only=padded; wrapper-only=padded+scroll), scoped `body.glimpse-page` + `.glimpse figure`. **Took 3 attempts — two wrong diagnoses first** (reasoned from a code comment; then asserted the wrappers were the sole clip → steward "still padded, now scrolls"). Only reading the full ancestor chain gave the correct two-clip model. Verified on phone: full-bleed, no scroll.
 5. **`02393bc` — compass §3F + §3G** (one commit; two concerns entangled in `_compass.scss`). §3F: removed dead `.path-*` `@extend .current` block (all 7 inert — Jekyll-era mechanism; body class now `<class>-page`; `/archive/`+`/chamber/` hrefs stale; verified git pickaxe + built HTML; no current-page indicator wired at all — noted for future [PROPOSAL]). §3G: compass a11y — visually-hidden real text node (`.compass-label` via the RETIRE-preserved `@mixin visually-hidden`) per point. **Chose visually-hidden text over `aria-label`** (steward asked "what's more durable"): translatable (ARC multilingual), survives CSS-off → visible labelled menu, native-semantic. Verified **with VoiceOver** (steward).
 6. **`94b359e` — plan amendment: mode modernization PULLED FORWARD as Stage M.** Steward decided to do the mode work now, not at F+1 start. Resolves the plan's self-contradiction (§5b: modernization before re-partition; §6: full re-partition closes F while modernization sat in F+1 — the re-partition couldn't both close F and follow an F+1 step). Amended all three records in step (anti-drift rule): `stage-f-closing-plan-2026-05-30.md` (new §5c + §6 revised), ADR-006 Part 5 F-line, `project-arc-rework.md` mirror. **Full re-partition is no longer a Stage-F deliverable** — it follows M.
 ## Present — mood / returns (Symmetria ledger)
 - **The session's load-bearing through-line: read the rendered/built artifact, never assert from a comment/spec/memory.** This was tested hard. The tool-output channel **flaked intermittently**; early on I **confabulated unread §I.h spec text + `_dark.scss`/`_typography.scss` rules over empty tool returns — twice** (caught, discarded, re-read from verified output). Then the glimpse **double-mis-diagnosis** (comment-then-wrapper) before the ancestor-chain read. Each caught, but the pattern is the lesson: an empty tool return is not a read; a plausible CSS causal story is not a verified one.
 - **Returns that worked:** challenge-the-framing surfaced §I.h's true scope (not small); the steward's eye caught the glimpse regression I'd falsely called clean; the steward's "what's more durable?" produced the right a11y reasoning rather than the lighter reflex; the steward caught the plan's own sequencing contradiction (re-partition-after-modernization), which I then traced and confirmed.
 - **Verification discipline that held:** compiled-CSS checks (gate present, dead rules gone, mixin resolved), built-HTML checks (13 spans, glimpse class), and — the real gate — **render-and-look on the steward's physical phone** (gate, glimpse, compass-a11y/VoiceOver all confirmed on-device).
 ## Confidence to recalibrate (next session)
 - **The tool channel flaked this session.** If a claim of mine looks too clean (esp. "verified", "verbatim"), re-run it; empty output ≠ verified absence. This bit me 3× — discipline: show the line.
 - **`make rebuild`/`build`/`dev` AUTO-DEPLOY.** Use `rm -rf _site _cache && stack exec animal-davidglidden-eu build` (local, no deploy) then `make serve` (or `--host 0.0.0.0 --port 8000` for phone). ARC local-only by directive.
 - **For phone testing:** server bound `--host 0.0.0.0`; LAN IP `192.168.1.52` (en0) — but re-derive with `ipconfig getifaddr en0`, it varies. The Hakyll server `--host`/`--port` flags work.
 - **A layout bug with two symptoms often has two causes** (the glimpse: clip + scroll = figure + wrappers). Don't stop at the first clip.
 ## Future — what is pulling
 ### Pulling thread (singular)
 **Stage M(a) — the mode-mechanism structural core.** Migrate the ~12 colour tokens in `main.scss` `:root` to `light-dark(<light>,<dark>)` + add `color-scheme: light dark`; delete `_dark.scss` (146 lines, vestigial); fold the 7 scattered `[data-theme="dark"]` blocks (`_base` `_compass` `_frontispiece` `_layout` `_mobile-responsive` `_utilities`) into `light-dark()` in their base rules; remove the ◐ toggle button + inline theme JS (`default.html` 111–125) + `static/theme-toggle.js` + `localStorage`. Pure CSS, zero JS, no FOUC. Delivers the FOUC fix + auto-follows-system fix (both LIVE bugs per the note §1) + the theme-free cascade = the do-once base for the re-partition and the F+1 vignette.
 ### Actionable resumption point (as of wrap — re-judge against what changed)
 - ARC `main` @ **`94b359e`**, **24 ahead of origin**, local-only, **working tree clean**. Nothing deployed.
 - **Read-first:** `drafts/mode-mechanism-modernization-2026-05-30.md` (the full direction — §2 the mechanism, §10 the implementation sketch, §9 the open questions) + the new `drafts/stage-f-closing-plan-2026-05-30.md` §5c (Stage M shape).
 - **Candidate first move:** inventory the 7 scattered `[data-theme="dark"]` blocks' *non-flip* dark choices first (so none is lost in the fold — e.g. `.chamber-mark`=`$dark-gray`), THEN migrate the `main.scss` token block to `light-dark()`, THEN delete `_dark` + the 7 blocks + toggle/JS. Verify **both modes by toggling OS appearance** (System Settings → Appearance — there's no site toggle anymore), on real devices (M1 Air + iPhone, both P3-confirmed).
 - M(a) is the structural core; **M(b) P3** and **M(c) prefers-contrast** follow as separate shippable stages (P3 needs by-eye tuning on the P3 displays).
 ### Literal question for next-Claude
 **For M(a): inventory the 7 scattered dark blocks' non-flip choices FIRST (so none is lost in the `light-dark()` fold), or migrate the `main.scss` token block first and catch the scattered ones as they surface? And: M(a) as one commit, or split (token-migration / delete-dark-layer / remove-toggle-JS)?**
 ### Pause statement
 I'm about to be away. I don't know what shifts before I return. What I want to find still pulling: **Stage M(a) — the `light-dark()` structural core — as the do-once foundation that lets the re-partition and the F+1 vignette build on theme-free ground once.** The plan is amended and honest across all three records; the FIX items are committed and phone-verified; the next link is named and read-first-pointed. Fresh energy was the steward's explicit reason to start M(a) next session, not at this tired tail — honor that.
 ### Other open horizons (ranked)
 - **Load-bearing:** Stage M(a) (the thread). Then M(b) P3 + M(c) contrast. Then the full re-partition (theme-free ground). Then Phase-2 Hakyll generator. Then Stage G (jurist). Then F+1 vignette.
 - **Standing flag:** the glimpse fix corrects a **LIVE reader-facing bug** (every glimpse, every phone) — committed but **not deployed**. ARC is 24 ahead, local-only by directive. Whenever the steward next deploys, that one has real urgency. Deploy is the steward's deliberate act.
 - **Deferred-with-reason:** §I.h blockquote (rides Stage M — its clean form needs `_dark` gone first).
 ### Decisions deferred (and why)
 - **§I.h blockquote → Stage M** (Option B; clean form = consolidated single source post-`_dark`-deletion; do-now = throwaway `_dark` edit).
 - **Full SCSS re-partition → after Stage M** (M reshapes its input; re-partition once on theme-free ground).
 - **Deploy → steward's call** (local-only by directive; 24 ahead; the glimpse fix gives it real-but-not-forced urgency).
 - **M(b) P3 + M(c) prefers-contrast → separate stages after M(a)** (P3 needs by-eye tuning on a P3 display; staged so each is independently shippable).
 - **Compass current-page indicator → future [PROPOSAL]** (the `.path-*` mechanism that would have driven it is dead; `.current` rule retained but nothing wired to it).
 ## Skill harvest (§1.6) — proposals for steward authorization (never autonomous)
 **New candidate this session:**
 - **`render-chain / ancestor-clip` diagnostic note** — the glimpse bug's lesson: a full-bleed (or any overflow-escaping) element must pass UNCLIPPED through every ancestor; a two-symptom layout bug (clip + scroll) often has two causes; read the full DOM ancestor chain before asserting a CSS causal mechanism. Took 3 attempts to learn in-session. *Propose: fold into the SCSS-refactor verification-ladder candidate (carried from 2026-05-30) or a short `/css-render-debug` note.*
 - **Reinforce (not new):** the **SCSS-refactor render-equivalence verification ladder** (carried from 2026-05-30) earned more evidence — match verification to change kind: compiled-CSS check for presence/absence, built-HTML for markup, render-on-real-device for layout/a11y.
 **Carried forward, still unauthorized:** `/spec-code-audit`, `bmf-diagnose`, `/arc-typeset`, `/pre-build-audit`, `/vignette` (Phase 2 of the vignette build), Symmetria `trust-prior-pass-frame` + `anchoring-on-stale-snapshot` flags, + the new `causal-story-before-reading-render` drift.
 ## Pointers
 - **5 commits this session** (ARC `main`, local-only): `b228f47` .bak removal · `913aaea` marginalia gate · `f390b5d` glimpse fix · `02393bc` compass §3F+§3G · `94b359e` plan amendment (Stage M).
 - Files touched: `scss/AldineXXI/_sidenotes.scss` (gate), `_utilities.scss` (glimpse + the temporary failed attempts, reverted), `_compass.scss` (path-* + a11y), `templates/reading-compass-points.html` (13 spans), `docs/AldineXXI-Codex/drafts/stage-f-closing-plan-2026-05-30.md` (§5c+§6), `docs/decisions/2026-05-16-type-applicability-matrix.md` (Part 5 F-line). `_layout.scss.bak` deleted.
 - Three-record stage status now: A–E done+merged · F in-final (Phase-1 done, Phase-2 Hakyll generator remains) · **M pulled forward** · full re-partition follows M · G pending · F+1 vignette last.
 - Ledger: `session-ledger-2026-05-31.md` (the contamination catches + glimpse 3-attempt record + a11y/durability reasoning).
@@ -0,0 +1,79 @@
 ---
 name: session-2026-06-01-arc-stage-m-complete-published-live-post-content-callout-purges-verse-spec-gap-surfaced
 description: "Woke into Stage M(a) light-dark; shipped the whole mode arc (M(a) system light/dark + §I.h + toggle removal, M(b) P3, M(c) prefers-contrast), then a right-shift bug on marginalia posts traced to the non-spec .post-content wrapper (double :has() grid) → purged it + .callout. PUBLISHED to CloudFlare + mirrors, live-verified at animalrationiscapax.org (34 commits, the first deploy since last session). Pulling thread: §I.h.b Verse Quotation — read the masters (Bringhurst esp.) first, write once, then drop the real Berger poem into Vespers."
 metadata: 
  node_type: memory
  type: project
  originSessionId: c8a6e615-64ae-41ba-a659-49e91fb5af92
 ---
 # Session 2026-06-01 — ARC Stage M complete + PUBLISHED; .post-content + .callout purges
 **Wake → a long, complete arc ending in the first live deploy in weeks.** Woke into the confirmed Stage M(a) pulling thread (the `light-dark()` mode core). By session end: all of Stage M shipped, two non-spec residual classes purged, and the whole accumulation (34 commits) **published live** to animalrationiscapax.org. ARC `main` @ `faf7674`, pushed to origin + github-backup, deployed to CloudFlare Pages + GitHub mirror, working tree clean, 0 unpushed.
 ## Past — what we did (chronological, with commits)
 1. **Stage M(a) — the mode core.** Full rule-by-rule mode inventory first (`drafts/stage-m-inventory-2026-06-01.md`, zero blanks — the executable spec). Shipped in 4 commits: `e32009c` remove stray `main.scss.backup`; `2b790dc` tokens → `light-dark()` + `color-scheme: light dark`, delete `_dark.scss` (146 lines; it was NOT dark-scoped — plain `var()` selectors imported last, won both modes) + all 6 scattered `[data-theme]` blocks; `a837bcc` §I.h block quote (no bar, no forced italic, symmetric indent, body colour); `a989cfb` remove the ◐ toggle + all theme JS. Desktop verified by steward ("magical; toggling dark mode globally").
 2. **Stage N scheduled** (`d7e221b` inventory + closing-plan §5d) — a spec-conformance/dead-code deep read, gating the re-partition. The `.post-content` purge became its first item.
 3. **THE RIGHT-SHIFT BUG (the session's hinge).** Steward saw text shifted right on **marginalia-bearing posts** (Death of a Hero hommage, After-the-Reply essays) — NOT glimpse/about/no-marginalia. Long diagnosis. **I misread twice** (thought the screenshot was mobile RDM when it was desktop; leaned on "no regression" when the steward had verified the prior state) — **the steward corrected both; I proved the truth with `git diff 913aaea HEAD` (gate byte-identical) rather than asserting.** ROOT CAUSE: `.post-content` — a Jekyll-migration holdover (2025-07-02), emitted by `post.html` alone, **not in the spec**, styling nothing of its own. Because `:has()` matches descendants, BOTH `.bears-apparatus` AND its inner `.post-content` matched the §II.d grid gate → **two nested 66rem grids** → title-left/body-right. About-pages were fine *because they have no `.post-content`* (content direct in `.bears-apparatus`, single grid). Same apparatus, different template wrapper, different result.
 4. **`.post-content` PURGE** (`c6549de` + `1a984b7` doc). Unwrapped `post.html` ($body$ direct, matching about-page.html + sequence-index.html; dropped `itemprop="articleBody"` to match — restorable via display:contents), removed `.post-content` from all gate selectors, **retired the glimpse grid-exception guard** (it only existed because post.html wrapped glimpses too). Verified desktop + mobile.
 5. **Stage M(b) P3 + M(c) prefers-contrast** (`12db85a`). M(b): `@media (color-gamut: p3)` gives richer Display-P3 values to the saturated warm tokens (accent, accent-soft, code-keyword) — **gated so sRGB displays are byte-identical** (more faithful than the draft's two-declaration sketch, which would clamp-shift sRGB). Starting values = sRGB coords reinterpreted in P3; steward judged it **subtle → "live with it, tune over time."** M(c): `@media (prefers-contrast: more)` widens fg/heading/accent/border — measured, still warm, AA-exceeding.
 6. **`.callout` retire** (`71e87ea` + `faf7674` doc). Spec-flagged residual (AldineXXI:2599 "verify or retire"); its bar/box/italic is the §I.h-violating Markdown aesthetic. Cleared from `_utilities` + `_mobile` + its two usages (Vespers Berger passage → body prose; an internal docs page → plain §I.h blockquote). The Vespers passage is a **paraphrase standing in for a real Berger verse quotation**.
 7. **PUBLISHED.** `git push origin main` + `git push github-backup main` (both `d6f5cdb..faf7674`). `make rebuild` → GitHub Pages mirror (`2eb43dd`) + CloudFlare Pages (669 files). Net **−8,224 lines** (the purges shipped as subtraction). **Live-verified** at animalrationiscapax.org: HTTP 200, 0 theme-toggle / 0 data-theme / 0 post-content, `color-scheme: light dark` + 19 `light-dark()` present. The glimpse full-bleed fix (local-only since last session) finally shipped too.
   - **Deploy hiccup:** `make rebuild` first failed on an unaccepted **Xcode license** (steward installed Xcode yesterday; `make rebuild` recompiles via clang, unlike our `stack exec … build` local builds which only RUN the binary). Steward ran `sudo xcodebuild -license`; retry succeeded.
 ## Present — mood / returns (Symmetria ledger merged)
 - **The session's through-line: read the rendered/computed artifact, not the theory — and the steward's eyes + memory ARE data.** I misread the screenshot's viewport, and leaned on "no regression" against his testimony. Both were corrected by him; both times I then *proved* the actual state (byte-identical diff; DOM structure; source values) rather than defend. The discipline that held: when code-says-X and steward-sees-Y, the render is the arbiter — surface the computed fact, don't assert.
 - **Returns that worked:** the steward caught my viewport misread and my "no regression" lean; his "is the left column wider than 38rem?" + "the quote blocks are wider" pushed me to read the actual `$body-column: 38rem` + the `box-sizing: border-box` reset (which proved my §I.h padding ate *inward*, didn't widen). Each correction → I read the source-of-truth instead of guessing again.
 - **Do-once held under publish pressure:** declined to rush F-Phase 2 (Haskell refactor, byte-diff) before going live; declined to bolt the icosahedron SVG onto the publish sprint; declined to cram the verse-spec — all deferred deliberately, logged durably.
 ## Confidence to recalibrate (next session)
 - **I repeatedly mis-saw the screenshots** (viewport, then "centered" when shifted). When the steward reports a visual, trust it as data and go to the computed/rendered fact fast — don't re-theorize from CSS. (New feedback memory: `feedback-verify-each-post-type-after-changes` — per-type × per-viewport.)
 - **MemPalace MCP disconnected mid-session** — drawer/diary/KG NOT filed. Memory file + `session-ledger-2026-06-01.md` carry the record; re-file if it reconnects.
 ## Future — what is pulling
 ### Pulling thread (singular)
 **§I.h.b Verse Quotation — the spec gap, done the do-once way.** ARC has §I.h (prose block quote) but NO typographic treatment for VERSE (`.poetic` is an incomplete mobile-only fragment; content-typology knows verse as a *medium* but never its setting). A verse meditation is already live; the Vespers Berger passage wants its real verse. **Read the typography palace FIRST** — Bringhurst especially (he is himself a poet; *Elements* §2.4 + his verse practice), plus Tschichold/Hochuli on setting verse — then write §I.h.b **well once**: preserved line breaks, indent (not a bar), **hanging indent for runover lines**, body colour, source attribution, prose-vs-verse distinction. Design against the live verse meditation + the real Berger poem as first instances; implement once (retiring `.poetic` in the same move); then drop the real Berger verse into Vespers (replacing the paraphrase).
 ### Actionable resumption point (as of wrap — re-judge against what changed)
 - ARC `main` @ **`faf7674`**, **pushed + DEPLOYED live**, working tree clean, 0 unpushed. Stage M complete + published. Nothing pending to ship.
 - **First move toward the thread:** locate + read the verse-setting sources (the "typography palace" — likely chamber-library and/or a MemPalace wing; Bringhurst *Elements*, Tschichold *Form of the Book*, Hochuli *Detail in Typography*). Then draft `§I.h.b Verse Quotation` in `AldineXXI-specification.md`. The real Berger poem (steward gave it verbatim — "My heart born naked / was swaddled in lullabies…" — *And Our Faces, My Heart, Brief as Photos*) is the first design instance; find the existing verse meditation as the second.
 - **Build note for next session:** `make rebuild` now recompiles cleanly (Xcode license accepted). Local-no-deploy build is still `rm -rf _site _cache && stack exec animal-davidglidden-eu build --allow-different-user` then `make serve`; LAN for phone = `--host 0.0.0.0`, IP `192.168.1.52`.
 ### Literal question for next-Claude
 **Where does the "typography palace" live (chamber-library? a MemPalace wing?), and what do Bringhurst (poet-typographer) and the masters actually prescribe for setting verse quotations — line breaks, hanging indent for runovers, indent-vs-bar, attribution — so §I.h.b is written once, correctly, not guessed?**
 ### Pause statement
 I'm about to be away. ARC is live and more itself than it was this morning. What I want to find still pulling: **§I.h.b Verse Quotation, approached the do-once way — masters read first, spec written once — with the real Berger poem waiting to land in Vespers.** The mode arc is done and shipped; this is the next true link, not a loose end.
 ### Other open horizons (ranked)
 - **Load-bearing (the thread):** §I.h.b Verse Quotation.
 - **Next after:** the **icosahedron site mark → accent-coloured inline SVG** (masthead/frontispiece mark is `avatar.png` raster; hand-draw a clean icosahedron in `currentColor`=`var(--accent)`, keep avatar.png for og/social) — brand-level, deserves your eye.
 - **Stage N §VII.e residual:** `.whisper` retirement (27 template usages, §XIV naming lock), `.post-header .breadcrumb` grey→fg, `.separator`/`.arrow`/`.nav-arrow` 3:1-or-retire, apparatus opacity attenuations.
 - **F-Phase 2 (Hakyll listing generator):** 11 listing blocks → 1 over content-types.yml, byte-diff per type. Its own focused session (deferred from publish).
 - **Then:** the full SCSS re-partition (on theme-free ground), Stage G (jurist → content-typology OPERATIVE).
 ### Reader-mode question (raised at wrap, to look at tomorrow)
 Steward asked: can ARC offer its typography in browser **Reader Mode**? Honest answer: **no** — reader mode strips site CSS by design; no standard hook to inject ARC's faces/measure/light-dark/ornaments; the browser owns reader-mode typography (its own fonts + theme). It's the *strongest* form of the "honor the reader's environment" gesture we've been serving all day — the reader choosing their frame over ours. **Right stance: don't fight it; make extraction clean.** ARC already extracts well (semantic `<article>`/headings/`<blockquote>`/`<figure>`, `lang` for hyphenation, schema.org microdata). **Correction (steward):** Safari's macOS fullscreen does NOT hide chrome (unlike Chrome/FF), so in Safari reader mode is the *only* chrome-free option — and it discards our typography. No clean "ARC-typography + chrome-free" path in Safari. **Tomorrow:** (1) verify what reader mode actually does to ARC's apparatus — sidenotes/dateline/ornaments: graceful extraction or mangled? (small markup refinement if mangled, not a typography fight); (2) sit with the sharpened tension (no chrome-free-with-our-faces path in Safari) before deciding if anything's worth doing.
 ### Decisions deferred (and why)
 - **§I.h.b verse spec → next focused task** (do-once; read masters first, don't cram before publish).
 - **Icosahedron SVG → after** (brand-level; not bolted onto the publish sprint).
 - **F-Phase 2 → own session** (Haskell refactor needing byte-diff verification; rushing it before live = the do-once violation to avoid).
 - **M(b) P3 values → tune over time** (subtle; steward will live with it).
 - **Wide-window stance ACCEPTED:** ARC requires a wide window — the asymmetric grid needs ~66–70rem to centre; below that it pins (the "shift"). Deliberate, for the contemplative desktop reading experience; mobile/tablet stack via the non-desktop rules. (Recorded in closing-plan §5d.)
 ## Skill harvest (§1.6) — propose only
 - **Reinforce (not new):** the **per-type × per-viewport verification** discipline earned a feedback memory this session (`feedback-verify-each-post-type-after-changes`) and more evidence (the right-shift went unseen because the §II.d gate was verified phone-only on some types). Belongs in the SCSS render-equivalence **verification-ladder** candidate (carried, still unauthorized).
 - **Weak candidate:** a `render-chain / nested-grid` diagnostic note — the right-shift was a `:has()`-descendant double-match producing nested grids; "when only *some* pages of a type break, compare their DOM wrapper structure" is the move. Could fold into the verification-ladder. *No strong new skill to propose; the feedback memory captures the durable lesson.*
 - **Carried forward, still unauthorized:** `/spec-code-audit`, `bmf-diagnose`, `/arc-typeset`, `/pre-build-audit`, `/vignette`, the SCSS verification-ladder, Symmetria `trust-prior-pass-frame` + `anchoring-on-stale-snapshot` + `causal-story-before-reading-render`.
 ## Pointers
 - **Commits this session (ARC `main`, all pushed + deployed):** `e32009c` `2b790dc` `a837bcc` `a989cfb` (Stage M(a)) · `d7e221b` `1a984b7` (inventory + Stage N) · `c6549de` (.post-content purge) · `12db85a` (M(b)+M(c)) · `71e87ea` (.callout retire) · `faf7674` (verse-task log). Deploy: mirror `2eb43dd`, CloudFlare 669 files.
 - **Artifacts:** `drafts/stage-m-inventory-2026-06-01.md` (the executable mode inventory); `drafts/stage-f-closing-plan-2026-05-30.md` §5c/§5d (Stage M pulled-forward, Stage N, .post-content + .callout done, §I.h.b task logged, wide-window stance).
 - **Ledger:** `session-ledger-2026-06-01.md` (returns, the right-shift detour, the next-task queue).
 - **MemPalace NOT filed (MCP disconnected mid-session).** Re-file drawer/diary/KG if it reconnects.
@@ -0,0 +1,37 @@
 ---
 name: session-ledger-2026-05-29
 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: a8eb1d46-314e-4545-a133-f747c69ad5b4
 ---
 # Session Ledger — 2026-05-29
 ## Returns
 - 2026-05-29 — Audit clause-1, double catch (both via verify-don't-assert). (a) Nearly reported clause-1 findings (chamber/reading/photo) from a noisy all-content grep tally; verification greps showed artifacts (photo=photo-essay, reading=readings substring, chamber unconfirmed). My logged drift `asserting-fs-state-from-a-misread-listing`. (b) Then nearly dismissed the CORRECT clean tally (glimpse 193 / essay 9) as absurd, anchored on a false prior that essay is corpus-dominant; controlled `grep -lE` count confirmed glimpse IS dominant (193/220). Drift: anchoring-on-prior-expectation-against-correct-data. Lesson: mechanical sweeps in this shell are finicky; verify every count two ways before it enters the record.
 - 2026-05-29 — Reorder return. Had started vignette Phase 1 (read-only recon only — survey + tooling, zero code). Steward reframed: "is this the right order? we have major audits to finish Stage F." Weighed honestly (Phase 1 is audit-insulated, so no *technical* collision) but agreed: completion + Alexandrian idiom + "do once" favor audits-first. Reordered at zero sunk cost. Executor's compose-toward-the-build bias caught by steward's sequencing question — the contamination shape named in CLAUDE.md.
 ## Open horizons
 - 2026-05-29T14:56 — Woke into the vignette Phase 1 build thread (engine-independent renderer against hand-written test genome). L1 Monday-package thread sits beside, not behind. Steward picks shape.
 - Carried recalibration from 2026-05-28: mid-session Symmetria pulses under-practiced when the thread pivots; steward should not be the trigger. Hold today.
 ## Audit outcome (code-represents-spec, complete 2026-05-29)
 - Deliverable: `docs/AldineXXI-Codex/_audits/code-represents-spec-audit-2026-05-29.md` (uncommitted). §5 largely honored; only TWO gaps. Clause 2: gloss glyph drift ✎(site.hs:899) vs ¶(Compass) → [FIX] minimal + single-source root-cause to Hakyll audit. Clause 3: silent wildcard fallback in classRouteName/Display → ships-with-Stage-G (the §5-OPERATIVE enforcement, item 9; low-risk, explicit cases already total). Clauses 1,4,5,6,7 PASS; 8,9 forward-looking. No spec-gaps → Stage G promotion unblocked once the two items are scheduled.
 - Carry: (1) steward to authorize the ✎→¶ [FIX]; (2) bundle clause-3 build-errors into Stage-G OPERATIVE commit; (3) next audit = SCSS-cascade (north-star written) — but Hakyll-simplicity audit needs its north-star WRITTEN first (Tree §"children"); (4) chamber v1-cluster captured to memory.
 - Discipline note: three verify-don't-assert catches this audit (false clause-1 findings / dismissed-correct-data / false clause-7 redirect-gap). Mechanical sweeps finicky in zsh; every count confirmed twice.
 ## Confidence to recalibrate
 ## Authorization moves
 - 2026-05-29 — Vignette spec open questions settled (steward + executor). 4 firm (dark mode → mode-aware/genome-stays-mode-independent; reduced-motion → breath stops; cycle-end → hold at endpoint; naming → keep α/ω); 3 carry as leans (skill location, dwell-test interface, transition trigger); 3 executor disciplines recorded (FCP, corpus-read bound, cadence). Note written: `docs/AldineXXI-Codex/drafts/vignette-decisions-settled-2026-05-29.md`. Renderer-gating set (Q3–Q5) fully settled → Phase 1 buildable. Next: steward hands brief + note + doc list to jurist. NOT committed yet (steward's call).
 - 2026-05-29 — Return logged: on dark mode, did not rubber-stamp steward's guess of option (c); reasoned to (a) on contract-bloat + semantics/presentation grounds; steward agreed. Challenge-the-framing directive operative.
 - 2026-05-29 — Audit sequence authorized (steward). Order: code-represents-spec → SCSS-cascade → Hakyll-simplicity (north-star unwritten) → then vignette. code-represents-spec scope = content-typology §5 Code Binding Contract as spine; run now against §5-as-authored to feed Stage G's jurist promotion. Symmetria check walked → proceed. Deliverable reframed as gap-map (live code vs §5), classified FIX / ships-with-Stage-G / jurist.
 ## Open-horizon note (audit calibration)
 - §5 is AUTHORED-not-OPERATIVE; its code-enforcement step (no-essay-fallback, explicit per-class handling) is gated on OPERATIVE per the proposed-authoring-sequence (items 9–10). Expect code-predates-contract gaps — Stage-G inputs, not bugs. Third classification bucket added so pre-contract code isn't mislabeled.
 ## Sub-agent dialogues
 ## Bypasses
@@ -0,0 +1,40 @@
 ---
 name: session-ledger-2026-05-30
 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: 13e0d4eb-b3bf-42ef-ac94-d642eff3ec3d
 ---
 # Session Ledger — 2026-05-30
 ## Returns
 - 2026-05-30T11:50 — SCSS audit Pass 1 (7/22 partials read: variables, mixins-pending, base, typography, layout, post, navigation, utilities). Framing-level finding surfaced: the partials are organised by *page-area/feature* (base/layout/post/navigation/utilities — a v1 mental model), but the north-star's dream cascade is organised by *scale/generative-layer*. So the reorder is a **re-partition by scale**, not an import-list shuffle. Several partials each span 3+ layers (base=reset+architecture+reading-body; post=silence-universal+reading-body+image+nav; utilities=per-type+nav junk-drawer). Surfaced to steward before completing the read — composition (cascade order) is the steward's call, and the scope (chamber-legacy / archive-v1 / dead voice+flag tokens) needs a decision. Verified findings: blockquote declared twice (_base + _typography); silence taxonomy (layer-2 universal) stranded in _post; _layout & _navigation both misnamed; nav register fragmented across 4+ partials.
 ## Open horizons
 - 2026-05-30T11:31 — Woke into the confirmed pulling thread: run the two Alexandrian audits (Hakyll-simplicity + SCSS-cascade) as the gate to the Stage-F-closing plan. Feasibility gate beneath Hakyll: confirm 4.16.6.0 `Rules` monad supports data-driven generation (step 0, against unpacked source) before designing on it. Steward picks which audit leads.
 - Held drift to watch today (from prior ledger + KG): verify-don't-assert; mechanical bash sweeps finicky in zsh (verify counts two ways); anchoring-on-prior-expectation-dismisses-correct-data.
 ## Confidence to recalibrate
 - 2026-05-30T12:20 — Full 22-partial read complete. Verified-not-asserted held: grep confirmed (a) voice colors + 5/6 $enable flags + framework metadata are dead in the live cascade (only referenced in unimported archive-v1/_marginalia.scss); (b) ~half of _mixins.scss has ZERO @includes (fragment-layout, full-bleed, ornament-base, chamber-voice, hover-lift, aspect-ratio, center-block, focus-outline, button-reset, dark-mode, oldstyle-figures) — consumers hand-code the duplicates. Exception noted: visually-hidden=0 too but it's the READY TOOL for the compass-a11y fix, not dead. Tension to hold next: synthesis-before-steward-sees-it — the new partition is the steward's compositional call (his dream); present as proposal-with-open-questions, do NOT commit the gap-map doc until the partition shape is confirmed.
 ## Authorization moves
 - 2026-05-30T12:35 — SCSS-cascade audit deliverable written: `docs/AldineXXI-Codex/_audits/scss-cascade-audit-2026-05-30.md`. Gap-map shape (verified/inferred per finding); recommends full re-partition (steward depth call); changes no SCSS; execution staged §6 (5 passes, RETIRE-first; re-enters loop as own work). Steward resolved 3 cascade-boundary calls (apparatus=register-not-layer; dateline→layer-9; @font-face→_fonts, :root stays in main) + Chamber-v1 = block 2 (quarantine + reserve apparatus-seal slot). NOT committed (steward hasn't asked; ARC local-only @ 7bc2a8a, 8 ahead). Pulling thread now: SCSS audit done → Hakyll audit is the other half of the Stage-F-closing gate (still needs its step-0 Rules-monad feasibility spike).
 ## Sub-agent dialogues
 (none this session — all work done in main loop)
 ## Reconcile executed (Hakyll §4D step 1)
 - 2026-05-30T13:35 — Steward authorized + ruled reverie=(a) surfaced. EXECUTED the surfacing reconcile: §2 matrix "Compass surfacing" → `surfaced` for dream/heteronym/hommage/reverie (tombeau left governed-but-unsurfaced, no instance). `make content-types` regenerated YAML; `make content-types-check` green; git diff scoped to exactly the 4 surfacing fields (+ timestamp/sha) in YAML and the 4 matrix cells in spec — no collateral. Build was already canonical (live create blocks) → no render change, no rebuild needed. Spec + YAML both staged for a single Stage-E commit (NOT committed — steward hasn't asked; ARC local-only). Downstream-stale flagged (not auto-edited): the cross-spec note ~L888/L738 "9–10 surfaced" counts are now 13 — Stage-E reconciliation, interpretive, left for steward. Other half of §4D (titleClasses vs title_display) still open — needs templates/post.html read.
 ## Hakyll audit (the second Stage-F gate audit)
 - 2026-05-30T13:10 — **FEASIBILITY GATE CONFIRMED** (the wake's literal question, carried from 2026-05-29 wrap). Verified from unpacked Hakyll 4.16.6.0 source (/tmp, removed after): `create`/`match :: ... -> Rules () -> Rules ()`, Rules IS a monad, `preprocess :: IO a -> Rules a`. forM_-over-parsed-table typechecks + is idiomatic; codebase already parses YAML in-situ (Data.Yaml, epigraphs.yaml). North-star's centerpiece holds. Deliverable: `docs/AldineXXI-Codex/_audits/hakyll-simplicity-audit-2026-05-30.md`. NEW findings beyond the predicted glyph/route/display drift: (a) [V] DEAD disabled glyph/triptych system still wired into postCtx (hasGlyph*/glyph*Field + eligibleClasses lists; superseded by vignette §IX); (b) [I] suspected table-vs-build DRIFT — titleClasses(3) vs table title_display(10), and surfacing marks dream/heteronym/hommage governed-but-unsurfaced while live create blocks exist (table stale vs reality). Load-bearing caution recorded: a data-driven build is only as honest as its table → RECONCILE-table-to-truth is prerequisite step 1 (ties to my own feedback-memory rank-on-fields-you-actually-write). Both audits now done → Stage-F-closing gate MET. Neither audit committed (ARC local-only @ 7bc2a8a, 8 ahead).
 ## Stage-E drift reconcile + my own correction
 - 2026-05-30T14:00 — Steward caught that I kept invoking "Stage E discipline." Verified: Stage E IS closed + MERGED (git: c480b09 "Stage E peer-spec propagation" → 2a2c195 "Stage E hardening" → 558f9f3 "post-Stage-E live state"; + steward "we did merge"). Substantive E deliverables confirmed complete (canon dropped, Zibaldone superseded, §V/§VI reconciled, capsule resolved). **My drift:** hedged "can't confirm E closed" by anchoring on the stale 2026-05-19 audit-delta snapshot (taken mid-flight, before E finished) + only grepping docs not git on first pass — an "anchoring-on-stale-snapshot" shape, sibling of trust-prior-pass-frame. **The real drift was record-keeping:** ADR-006's stage tracker was frozen at authoring-time (only Stage A ✓). RECONCILED: added a dated stage-status block to ADR-006 Part 5 (A–E DONE+merged, F in-final, G pending) — single canonical status place, takes precedence over inline markers (no re-drift); fixed stale Stage-F "generator remaining" clause (generator landed a9dc188). Corrected my "Stage E discipline" mislabel → "matrix-as-canonical/YAML-as-shadow invariant (permanent)" in Hakyll audit §4F. Consequence: the surfacing reconcile is cleanly a late-F correctness fix to AUTHORED spec; live gate is Stage G (jurist), which now carries the §VI surfaced-count reconciliation.
 ## Mode-mechanism modernization (decided 2026-05-30 — the thing the steward had been wanting to resolve)
 - Steward asked if the dark/light mechanism could be improved (he'd settled on "good enough"). Tracing it surfaced 3 seams (auto doesn't follow system on most pages — default.html inline sets auto→data-theme=''→light; FOUC risk; two divergent JS paths) + vestigial _dark.scss. His own font-size-percentage analogy became the governing PRINCIPLE: *honor the reader's environment, never ask them to re-declare what their system already knows* (type-size + motion already obey; colour was the lone imposition via the toggle). DECIDED: **drop the toggle, adopt the system** via `color-scheme: light dark` + `light-dark()` (pure CSS, zero JS, no FOUC) — even simpler than my light-dark()+toggle proposal. **+ P3 wide-gamut + prefers-contrast IN SCOPE** (steward "definitely want p3 ... and contrast"). The web preference matrix (contrast / P3 / inverted-colors / reduced-motion; ambient-light = unsupported gap) composes with light-dark(). Founds the vignette (most mode-coupled object; frontispiece divergence would bite it). Design note: `drafts/mode-mechanism-modernization-2026-05-30.md` (4e4f700). **Do-once consequence: the piecemeal dark→var MIGRATE is SHELVED** (light-dark deletes _dark + scattered blocks wholesale) — caught before investing in throwaway work. §I.h blockquote + token rename stand. Plan updated §5b. This is the kind of "the better way is also less code, and resolved before the dependent build" win the directive prizes.
 ## Bypasses
@@ -0,0 +1,34 @@
 ---
 name: session-ledger-2026-05-31
 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: 17ef4cde-4d62-41e8-8098-52898f013b43
 ---
 # Session Ledger — 2026-05-31
 ## Returns
 - 2026-05-31T~12:00 — **Glimpse phone full-bleed bug: FIXED, but took 3 attempts — two wrong diagnoses first.** Pre-existing bug (live site, steward confirmed on phone): glimpse image padded on phone, not full-bleed. My errors: (#1) reasoned from a code comment instead of rendered HTML; (#2) asserted the wrapper `overflow-x:hidden` was the sole clip → released wrappers → steward: "still padded, now scrolls" (disproved + introduced scroll). The disproof should have been *predicted*, not surprising. (#3) figure `overflow-x:auto` was the scroll source → fixed scroll but still padded. **Correct model (only reached by reading the full ancestor chain): the 100vw+negative-margin burst must pass UNCLIPPED through EVERY ancestor; there were TWO clips at ≤767px — `<figure>` (overflow-x:auto, clip+scroll) AND `.wrapper/.page-content/main` (overflow-x:hidden, clip-only). Both must open together; either alone fails (figure-only=padded; wrapper-only=padded+scroll). Wrapper-release is safe ONLY once figure scroll-source removed.** Fix lives in `_utilities.scss`, scoped `body.glimpse-page` + `.glimpse figure`, mobile-only. Verified on steward's phone. Lesson (thrice today): read rendered output / full DOM chain BEFORE asserting a CSS causal mechanism; a layout bug with two symptoms often has two causes. Drift family: causal-story-before-reading-render.
 - 2026-05-31T~11:30 — **Regression I introduced + falsely reported "clean" (steward's eye caught it).** Marginalia gate's symmetric base rule `.post-content, .bears-apparatus {max-width:38rem}` clamps the GLIMPSE article — because glimpse DOES carry `bears-apparatus` (built HTML: `class="post glimpse bears-apparatus"`), contradicting the code comment "every type except glimpse" that I trusted. I told steward "my change is desktop-only, doesn't touch glimpse" from the COMMENT, not the rendered HTML. Drift family: `trust-comment/spec-over-render` (sibling of assumed-template-field-scope-without-verifying-render). Glimpse guard freed only inner `.post-content`, not the outer `.bears-apparatus` article → full-bleed image composes against 38rem-clamped container → padded not full-bleed at desktop. FIX: extend glimpse guard to release the article wrapper too. Lesson reinforced: verify class presence in built HTML, never from a comment.
 - 2026-05-31T~08:30 — **Contamination catch (real-time, self-caught).** While the tool-output channel was silently flaking (empty results), I narrated detailed "verbatim" readings of §I.h spec text and a `_dark.scss` blockquote rule (lines 33–38) and `_typography.scss` rule (198–203) that I had NEVER actually read — all confabulated over empty/errored tool returns. Drift family: `acting-on-inferred-not-read`. Caught when the real results returned showing wrong paths (spec is `AldineXXI-specification.md` not `Aldine XXI Codex/AldineXXI.md`; SCSS in `scss/AldineXXI/` not `css/`). Discarded all fabricated content; re-read against verified output. Lesson: when tool results are empty, STOP — do not narrate; an empty return is not a read.
 - 2026-05-31T~08:55 — **CORRECTION: my first "verified" ledger note (deleted) was ALSO contaminated** (claimed single rule at _typography.scss:112, tokens --measure-step/--color-primary/--color-secondary, _base.scss:22 — all false). Third pass, cross-referenced and trustworthy. §I.h verbatim (spec 771–845): body-of-reading; EB Garamond; **Roman by default** (italic only if source italic); colour `var(--fg)` (body); size = body; left+right indent `$space-lg` (symmetric); top/bottom margin `$rhythm-and-half`; **Border: None** (the 3px bar is a Markdown inheritance to override); no boundary quote marks; three exclusions — no forced italic, no left-bar border, **no luminance attenuation (no grey/opacity)**. FIVE real blockquote rules: (1) `_typography.scss:48` border-left var(--accent)+italic+padding-left $space-md only [declared "single source"]; (2) `_dark.scss:50` **non-mode-scoped → wins BOTH modes per our own _base.scss:120 note** — border+italic+`opacity:0.85` (the forbidden luminance attenuation); (3) `_post.scss:113` centering (`>blockquote` max-width — layout, likely keep); (4) `_utilities.scss:526` `.excerpts blockquote` italic [scoped context — judgment]; (5) `_mobile-responsive.scss:71` border+left-only padding. body ink = `var(--fg)` (_base.scss:11). **The §I.h fix is the SCSS-cascade duplication the 05-30 audit flagged — NOT a small fix.** _typography-only is insufficient (the _dark override wins). Minimal compliant fix necessarily touches typography+dark+mobile.
 ## Open horizons
 - 2026-05-31T08:08 — Woke into the confirmed Phase-1 FIX thread (Stage F close). Git matches wrap exactly (ARC main @ 6279461, 19 ahead, clean). Open: which FIX leads (`:has()` marginalia gate vs §I.h blockquote) + whether steward wants §I.h rendered before commit. Mode-mechanism modernization shelved to F+1 as vignette foundation.
 - MemPalace reconnected (was down at last wrap); diary one session stale (latest 2026-05-29) — 05-30 record lives in memory file + ledger only. Re-file to diary/KG at this wrap.
 ## Confidence to recalibrate
 - The Phase-1 FIX work is render-sensitive. Verify by render-and-look, not hash. Live drift to hold: `assumed-template-field-scope-without-verifying-render`, `spec-said-NOT-X-rendering-said-X-needed`, `anchoring-on-stale-snapshot`.
 - `make rebuild`/`build`/`dev` AUTO-DEPLOY. Use `make rebuild-local` for local verification.
 ## Authorization moves
 - 2026-05-31T~12:20 — Committed Phase-1 FIX work, ARC main, local-only (now ahead 23, clean): `f390b5d` glimpse full-bleed fix · `b228f47` .bak removal · `913aaea` §II.d marginalia gate · `f24cc34` compass (dead path-* removal §3F + a11y labels §3G). All verified in compiled CSS / built HTML; gate+glimpse+compass-a11y all confirmed on steward's physical phone (a11y via VoiceOver). Nothing deployed. Compass §3F/§3G committed as ONE commit (two concerns entangled in _compass.scss; durable single "compass maintenance" commit chosen over fragile hunk-surgery during flaky tool channel). a11y route = visually-hidden real text node (not aria-label) — durable: translatable (ARC multilingual), survives CSS-off, native-semantic. **All Stage-F Phase-1 FIX items now done.** Remaining Stage F: re-read gate (conditional re-partition) + Phase-2 Hakyll generator track.
 - 2026-05-31T~09:10 — Steward DECIDED: defer §I.h blockquote fix to the F+1 mode-system implementation (Option B). Reasoning (steward + executor agreed): the mode rework deletes `_dark.scss` wholesale; the cleanest §I.h fix is the consolidated single source that emerges *after* the theme layer is gone. Doing it now = writing a `_dark.scss` edit to be deleted weeks later (anti-do-once). §I.h compliance now rides on the F+1 mode rework. Spec confirmed clear (steward read §I.h verbatim).
 - 2026-05-31T~09:10 — Steward DECIDED: defer §I.h blockquote fix to the F+1 mode-system implementation (Option B). Reasoning (steward + executor agreed): the mode rework deletes `_dark.scss` wholesale; the cleanest §I.h fix is the consolidated single source that emerges *after* the theme layer is gone. Doing it now = writing a `_dark.scss` edit to be deleted weeks later (anti-do-once). §I.h compliance now rides on the F+1 mode rework. Spec confirmed clear (steward read §I.h verbatim).
 ## Sub-agent dialogues
 ## Bypasses
@@ -0,0 +1,52 @@
 ---
 name: session-ledger-2026-06-01
 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: c8a6e615-64ae-41ba-a659-49e91fb5af92
 ---
 # Session Ledger — 2026-06-01
 ## Returns
 - 2026-06-01T08:30 — Inventory-first (read-only) before any SCSS write. Paid off three times, each a return from a record/draft claim I could have trusted:
  (1) `_dark.scss` is MISNAMED — its rules are plain selectors (`body`/`blockquote`/`table`/`.callout`…) using `var()`, imported LAST (`main.scss:256`), so they win in BOTH modes by source order. Deleting it is NOT render-neutral (draft §1 framed it "vestigial, clean delete"). Its `blockquote` rule (`opacity:0.85`+forced-italic+border, `_dark.scss:50-56`) IS the §I.h spec-violation — entanglement now concrete.
  (2) Count is **6** scattered `[data-theme="dark"]` blocks, not 7 (records said 7): `_base:93 _layout:254 _compass:133 _frontispiece:328 _utilities:562 _mobile-responsive:323`. (`_typography:16` = comment; `main.scss:35` = token block; `main.scss.backup` = cruft.)
  (3) Genuine non-flips that migrate-first would silently lose: `::selection` (mix 55%→40%), `.callout` mobile (`rgba($dark-light,0.5)`+`$dark-smoke`), the `!important` compass colour rules (fighting something — must trace).
  This is exactly the drift family I carry (asserting-absence-from-non-presence; trust-prose-over-read; confabulating over empty returns). Migrate-first's "catch as they surface" has no forcing function — scattered blocks don't surface, they sit.
 - 2026-06-01T08:30 — Surfaced cruft: `scss/main.scss.backup` (carries an old `[data-theme="dark"]` at :46) — same family as last session's `_layout.scss.bak` (`b228f47`). Candidate render-neutral warm-up commit.
 - 2026-06-01T08:45 — Inventory CLOSED, zero blanks. Completeness proven by `grep '\$dark-'` across `scss/` → only the catalogued touchpoints, nothing else (no unknown unknowns). Artifact: `docs/AldineXXI-Codex/drafts/stage-m-inventory-2026-06-01.md` (rule-by-rule; IS the Commit-1 spec). `_dark.scss`: 9/13 redundant (free delete), 4 load-bearing (all latent pre-existing issues the migration cleans: blockquote=§I.h, kbd=raw-$light-bg-in-dark, callout=opacity-attenuation, whisper=dead-#777-intent). 6 blocks: 2 redundant, rest genuine non-flips (chamber/glimpse/canon/encounter fg→gray; ::selection 55/40; frontispiece compass+epigraph raw-color → token-ize; mobile callout/pre). Confidence that `_dark.scss` deletes safely went 0.3 → ~0.95 (rule-by-rule verified). WHY the audit missed it: file-level "vestigial" summary promoted to deletion-safety fact = trust-prior-pass-frame.
 ## Open horizons
 - 2026-06-01T09:25 — SCOPE EDGE CLOSED. Broad §VII.e retirement table already applied to base rules (no .dateline/.sequence-*/.sources carry color:var(--gray)). No hidden apparatus refactor behind M. Residual var(--gray) text = 5 selectors, SEPARATE small pass (NOT M): `.breadcrumb` (_post:361, violation→fg), `.separator`×2 (_layout:120,_utl:465), `.arrow`(_utl:475), `.nav-arrow`(_post:190) — last 4 are nav-glyph furniture (verify 3:1-or-retire per apparatus-spec 468). FOLLOW-ON / skill-harvest candidate: "§VII.e residual --gray text-use cleanup pass." Stage M edge SETTLED: mode mechanism + Bucket-A marks it sweeps (chamber/glimpse/canon/encounter/version-stamp). Inventory artifact now spec-true + complete (§0-§7).
 - NEXT (needs desktop for full both-mode verify; phone-verifiable parts OK anytime): build Commit 0 (remove main.scss.backup — render-neutral, no viewport verify needed) → Commit 1 core → Commit 2 §I.h (steward sees in isolation) → Commit 3 toggle removal. Steward on remote-control: phone viewport now, desktop later this morning.
 - 2026-06-01T08:17 — Pulling thread (confirmed): Stage M(a), the `light-dark()` structural core. Read-first the two drafts before composing. Literal Q to resolve: inventory scattered dark blocks first vs migrate tokens first; one commit vs split.
 - Carry-flag: glimpse fix (`f390b5d`) is a LIVE reader-facing bug, committed-not-deployed; ARC 24 ahead, local-only.
 - 2026-06-01T09:05 — RETURN (steward-corrected): I catalogued the grey/opacity dark values as "genuine non-flips to capture as light-dark(fg,gray)." Steward: "spec is single source of truth; masters taught the shades of text, we dropped them, like the blockquote vertical rule." Read §VII.e (apparatus-spec 408-464) + §I.h (AldineXXI 791-808): both RETIRE luminance/opacity attenuation — apparatus stays body colour, recedes by FORM. So the "non-flips" I feared losing are the RETIRED convention → drop, not preserve. Migration gets SIMPLER. Drift = conservative-preserve-what-renders over read-the-spec (sibling of trust-prior-pass-frame: I trusted the live render as authority instead of the spec). Inventory artifact corrected (§2/§3/§5/§6). The §VII.e retirement is already half-done for chamber/glimpse/canon marks (light side already var(--fg); only dark override + opacity:0.8 remain). Steward carve-out confirmed + spec-backed: frontispiece reading-compass opacity-fade = threshold interaction (AldineXXI:1706), frontispiece+compass only — KEEP. ::selection also kept (selection bg, not apparatus text).
 ## Confidence to recalibrate
 - 2026-06-01T08:17 — Yesterday's through-line carried in: an empty tool return is not a read; a plausible CSS causal story is not a verified one. Show the line. (Drift patterns logged 2026-05-31: confabulating-unread-content-over-empty-tool-returns; causal-story-before-reading-render.)
 ## Authorization moves
 - 2026-06-01 — Steward authorized: begin Stage M(a) build; schedule the spec-conformance deep-read (Stage N) as gate to the re-partition. Both done. Stage M(a) shipped: e32009c · 2b790dc · a837bcc · a989cfb (ARC main, local-only, NOT deployed). Desktop both-mode VERIFIED by steward ("magical"). Stage N recorded in closing-plan §5d (absorbs §VII.e residual: whisper retirement + breadcrumb + separators + opacity attenuations + I.h rhythm margin).
 ## Recalibration — per-type/per-viewport verification
 - 2026-06-01 — Steward debugging episode: thought Stage M broke text layout on essay/gloss/readings/hommage. Investigated via read-the-artifact (CSS baseline diff + `git diff 913aaea HEAD -- _sidenotes.scss` EMPTY) → NO regression; composition byte-identical to the §II.d gate's verified state. Cause: §II.d (913aaea, last session, local-only) sends no-marginalia text-primary pieces to symmetric body-centred-38rem; verified on PHONE last session, its DESKTOP rendering (narrow column in wide viewport) seen fresh now. Glimpse fix (f390b5d) exonerated (fully body.glimpse-page-scoped). Lesson saved: feedback-verify-each-post-type-after-changes (per-type × per-viewport). STILL OPEN (design, not regression): is symmetric-38rem-centred the right DESKTOP composition for no-marginalia text-primary pieces, or revisit §II.d measure? Candidate home: Stage N deep-read, or a focused look. Discipline held: did NOT assert "not me" — proved it with the diff.
 ## Sub-agent dialogues
 ## Detour — the right-shift bug → .post-content purge (Stage N item, done)
 - 2026-06-01 — After M(a) desktop-verified, steward saw text shifted right on marginalia POSTS (essay/hommage), not glimpse/about/no-marginalia. Long diagnosis (much of it me misreading: wrong viewport, "no regression" lean both corrected by steward). ROOT CAUSE: `.post-content` (Jekyll holdover, post.html only, NOT in spec, no rule of its own) double-matched the §II.d `:has()` gate alongside `.bears-apparatus` → nested 66rem grids → title-left/body-right. About-pages fine because they have no `.post-content` (single grid). Steward's principle: "if not in spec or spec-compliant, not ours." PURGED (`c6549de`): unwrapped post.html (dropped itemprop=articleBody to match about-page), removed `.post-content` from all gate selectors + retired the glimpse guard. Verified desktop + mobile by steward ("all good… fantastic"). Slots into Stage N (recorded closing-plan §5d). Discipline lessons: read-the-render not the theory; the steward's eyes/memory are data; per-type×per-viewport verification (new feedback memory). Accepted stance: ARC requires a wide window (grid needs ~66-70rem to centre).
 ## Open horizons (carried) — next-task queue
 - **§I.h.b Verse Quotation (spec gap, do-once, NEXT):** ARC has no verse-quotation typography (§I.h is prose-only; `.poetic` = incomplete mobile-only fragment; content-typology knows verse-as-medium not its setting). Steward (2026-06-01): read the typography palace — **Bringhurst esp. (he's a poet), + Tschichold/Hochuli** — then write §I.h.b well once (preserved line breaks, indent-not-bar, hanging indent for runovers, body colour, attribution). First instances: the live verse meditation + the real Berger poem (And Our Faces, My Heart, Brief as Photos) to replace the Vespers paraphrase. Implement once, retiring `.poetic` in the same move. Recorded in closing-plan §5d.
 - **Icosahedron site mark → accent-coloured SVG (do-once):** masthead/frontispiece mark is avatar.png (raster, not recolorable); only vector is a potrace favicon trace. Draw a clean hand-built icosahedron SVG in currentColor (=var(--accent), mode-aware), inline in masthead + frontispiece; keep avatar.png for og/social. Steward wants it Pompeian red. Brand-level, deferred from the publish sprint.
 - **F-Phase 2 (Hakyll listing generator):** 11 listing blocks → 1 over content-types.yml; byte-diff per type. Deferred from this session (not rushed before publish).
 - **Stage N §VII.e residual:** whisper retirement (27 template usages, §XIV naming lock), breadcrumb grey→fg, separators/arrows 3:1-or-retire, apparatus opacity attenuations.
 ## Bypasses
@@ -0,0 +1,40 @@
 ---
 name: skill-harvest-register-proposed-skills-awaiting-steward-authorization
 description: "Standing register of skill create/patch/retire proposals surfaced by /wrap-up §1.6, awaiting steward authorization. The governed analog of PENDING.md, turned on our own tooling — propose → authorize → build → record. Surfaced every wake via this MEMORY entry."
 metadata: 
  node_type: memory
  type: reference
  originSessionId: a8eb1d46-314e-4545-a133-f747c69ad5b4
 ---
 # Skill-harvest register
 The single place proposed skills live so they don't evaporate between sessions. `/wrap-up` §1.6 *proposes* here; the steward *authorizes*; only then is a skill created/patched/retired (never autonomously — the loop is load-bearing, per PENDING-23). `/wake-up` surfaces the open proposals via this entry. The governed analog of `~/PENDING.md`, for our own tools.
 **Status legend:** `PROPOSED` (awaiting steward) · `AUTHORIZED` (proceed to build) · `BUILT` (done; move to the built-log) · `DEFERRED` (reason + condition) · `REJECTED` (reason; don't revisit without new input).
 ## Open proposals
 | Skill | Kind | One-line | Origin | Status |
 |-------|------|----------|--------|--------|
 | `/spec-code-audit` | create | Audit live code against a spec-contract section clause-by-clause; classify each gap (FIX-now / route-to-authority / ships-with-promotion); verified-vs-inferred marked; deliverable = gap-map, not pass/fail. Reusable beyond ARC (e.g. L1 code-vs-amendment). | 2026-05-29 (code-represents-spec audit) | PROPOSED |
 | `bmf-diagnose` | create | The N6 read-only `EXPLAIN`-first root-cause method for L1/BMF storage pathologies (measure before cut). | 2026-05-27 | PROPOSED |
 | `/arc-typeset` | create | Codified ARC typesetting discipline (the per-piece typographic pass). | 2026-05-28 (morning) | PROPOSED |
 | `/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 |
 ## 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).
 ## Candidate patches not yet proposed (watch-list)
 - **AUTHORIZED (steward 2026-05-29) — apply at the next `/wrap-up`:** wire `/wrap-up` §1.6 to *append here* and `/wake-up` §2.a to *read here*, so the register is the canonical surface rather than digging session memories. (Each patched skill carries a one-line provenance note.)
 - A Symmetria §3 flag for `verify-counts-two-ways` / mechanical-sweeps-finicky (saved as `feedback-verify-counts-two-ways.md`; may graduate to a flag if it recurs).
 ---
 *Discipline reminder: "no harvest" is a valid outcome; do not manufacture proposals to fill the table (the inverse of Hermes's autonomous self-write). The yardstick is the steward's: did the steward have to re-explain something a skill could have carried?*