dotfiles/claude/memory/skill-harvest-register.md

name, description, metadata, permalink

name	description	metadata	permalink
skill-harvest-register-proposed-skills-awaiting-steward-authorization	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.	node_type type originSessionId memory reference a8eb1d46-314e-4545-a133-f747c69ad5b4	claude-memory/skill-harvest-register

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

Full review with steward 2026-06-05 (housekeeping session) — every item ruled, evidence-checked against the KG drift/built record. Verdicts below.

Skill	Kind	One-line	Origin	Status
/spec-code-audit	create	Audit live code against a spec-contract section clause-by-clause; classify each gap; deliverable = gap-map.	2026-05-29	BUILT 2026-06-07 — deferral condition met from the ARC direction (Seb's completeness question revived the justification; steward authorized build-from-this-run). Generalizes the four-pass bidirectional method proven on the full ARC audit (_audits/spec-code-full-audit-2026-06-07.md: 172 findings, 66-agent fan-out, adversarial refutation, 2 kills). Skill at ~/.claude/skills/spec-code-audit/. Serves L1's first spec-vs-code audit as originally intended.
bmf-diagnose	create	N6 EXPLAIN-first root-cause method for L1/BMF storage pathologies.	2026-05-27	AUTHORIZED 2026-06-05 (build at next L1 diagnostic need).
mempalace-diagnose (was mcp-drop-diagnose)	create	Upstream-issue-first method for MemPalace/MCP failures (proved on #1495; HNSW quarantine + search errors 2026-06-05).	2026-06-02	AUTHORIZED 2026-06-05. Merge with bmf-diagnose RETRACTED same evening — the merge was itself the MemPalace≠BMF conflation (steward's 3rd correction): one substrate we build, one we merely use. Two skills, two names.
/pre-build-audit	create	The four-pass scope-claimed-vs-scope-tested per claim, with substrate evidence discipline.	2026-05-28	AUTHORIZED 2026-06-05, build-on-next-big-build (cascade Stage 2 / vignette Phase 1 / L1).
/vignette	create	Phase-α vignette genome generation under Symmetria.	2026-05-28	DEFERRED (until vignette Phase 2) — condition unchanged, re-confirmed 2026-06-05.

Built / authorized (lineage — for context, not action)

• /landscape-scan + /tooling-scan (the scan-skill family, per-workstream lens cards) — built + in use (2026-05-27).
• /wrap-up §1.6 skill-harvest step + /wake-up §2.a/§3 glance — the practice itself (PENDING-23, 2026-05-27).
• Chaîne-d'union clasp across wake / Symmetria / wrap (2026-05-26).
• /wrap-up §4.0 — MemPalace liveness-ping-before-diary; steward-authorized + applied (2026-06-02; #1495 cold-start drop resilience, so a silent drop never eats the wrap filing).
• /wrap-up §6.5 dotfiles session-state commit+push (scoped add, session-stamped message, preservation-not-modification boundary) + /wake-up §2.c read-only push-check — steward-proposed and authorized conversationally, built same session (2026-06-05 housekeeping). Closes the governance-state single-disk latency that sysupdate's calendar cadence left open; sysupdate keeps the catch-all sweep + Brewfile dump.
• Symmetria §3 — two flags added (2026-06-05, authorized + built same evening): trust-prior-pass-frame (frame-inheritance from prior verification into untested domain) + census-through-truncation (count first, then look). Closes the long-standing §3 patch proposal.
• reference-verification-ladder.md (2026-06-05, authorized + built same evening): one canonical home for the proven instruments (byte-identical compile gate, expected-delta/delta-classification, censused-routes, measure-toolchain-before-spec, compiled-selector grep, governed-char byte gate, WCAG compute, fresh-clone gate, live-fetch gate, per-type×per-viewport, re-verify-at-extension-scope). Ends the per-wrap re-proposing; future instruments append there, not here.
• /arc-typeset REJECTED (2026-06-05): superseded — §I.k.b/c moved the mechanical pass into the build; LINT class carries judgment warnings; multilingual §2.5 carries authoring notes; french-typography-pass covers the French preparation layer. Do not re-propose without a new gap.

Candidate patches not yet proposed (watch-list)

• Register wiring (authorized 2026-05-29) — APPLIED 2026-06-05 (evening wrap): wrap §1.6 now appends here; wake §2.a now reads here. Provenance notes in both skills.
• 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).

New proposals (2026-06-06 wrap — awaiting steward)

Skill	Kind	One-line	Origin	Status
bmf-diagnose	build now (already AUTHORIZED 2026-06-05 "at next L1 diagnostic need")	Today WAS that need and the method is proven+fresh: process sample → log pattern census (uniq -c histogram) → SIGUSR1→CDP CPU profile (working script at /tmp/profile-mindfabric.mjs — fold in before /tmp clears) → read-only SQLite census (?mode=ro URI) → evidence preservation to l1-reliability/evidence/ → restart-with-capture. Hand-derived today over ~2h; the skill carries it next time in minutes.	2026-06-06 mindfabric-00 forensics	PROPOSED (build, not authorize — authorization exists)

New proposals (2026-06-05 evening wrap — awaiting steward)

Skill	Kind	One-line	Origin	Status
/wrap-up §5	patch	Palace-fully-derived, part 1: mirror every kg_add/kg_invalidate made at wrap into the session memory file (one line each), so KG facts — the last palace-only data — become file-backed and the palace becomes wholly disposable/rebuildable.	2026-06-05 MemPalace forensics + tooling scan part (b)	PROPOSED
/wake-up §2.b	patch	Until upstream #1665 closes: wake searches run unscoped + post-filter by wing (wing-scoped mempalace_search errors at HEAD). Remove when the feedback memory feedback-mempalace-wing-filter-broken is deleted.	2026-06-05 diagnosis	PROPOSED
/wake-up (new step or §2 check)	patch	Wake canary: seconds-cheap probe at wake — every MEMORY.md pointer + [[link]] resolves to an existing memory file; flag dead pointers and load-bearing orphans. Lesson 7 of substrate-failure-lessons-2026-06-07 (BM false-clean sync) applied to our own index, which the harness already part-loads with a warning.	2026-06-07 evening (memory-verdicts night)	PROPOSED

New proposals (2026-06-08 — harvested from Fowler, Refactoring 2nd ed. chs 1–3, awaiting steward)

Read in the Chamber before Wave 3R. Three elements are genuinely NEW to our toolkit (not already carried by the verification ladder / Symmetria); two more are heuristics worth a feedback memory. Proposing the new ones; flagging where each would land on authorization.

Element	Kind	One-line	Where it lands	Status
Revert-and-redo-smaller	Symmetria §3 flag OR ladder note	Ch1's strongest working reflex we don't have: when a verification/gate fails and the cause isn't immediately visible, revert to the last green commit and redo the step smaller — do NOT debug forward. We default to debug-forward; this is a different reflex, and it depends on commit-after-every-green (which we already do via the byte-identical gate).	Symmetria §3 (a flag) or verification-ladder (a named discipline)	PROPOSED
Two-hat commit separation	verification-ladder entry	Name which hat each commit wears: a refactor commit's compiled output is byte-identical (or carries a pre-stated classified delta); a feature/behaviour commit is where rendered values change. We already own the byte-identical gate + delta-classification — this names the commit-level rule that pairs them ("never mix a structural move and a value change in one step"). The ARC @layer re-partition is the textbook case.	verification-ladder (formalizes what the byte-gate implies)	PROPOSED
Bad-smells → refactoring lens	reference card OR fold into /code-review	Ch3's smell catalogue (Mutable/Global Data, Duplicated Code, Shotgun Surgery, Speculative Generality, Comments-as-deodorant…) as an explicit review lens. Today it classified the !important census cleanly (global-data smell → @layer). Likely too large for a new skill; candidate to fold into /code-review's prompt or a one-page reference beside the ladder.	/code-review prompt or new reference	PROPOSED (lowest priority — may be redundant with existing review skill)

Heuristics (feedback-memory candidates, no skill needed):

• Rule of Three (ch2): first time do it, second wince, third refactor. Defer abstraction to the third repetition.
• Preparatory refactoring (ch2, Kent Beck): "make the change easy (warning: this may be hard), then make the easy change." The framing for sequencing Wave 3R before Stage G — Jessica Kerr's "drive north to the highway."

Note: measure-don't-speculate (ch2 date-range story) is NOT proposed — already carried by the ladder (censused-routes, measure-toolchain-before-spec). Today's census re-proved it; no new instrument needed.

New proposals (2026-06-08 evening — the tree + the memory-index archive pass)

Verification-ladder additions (the session proved both; → reference-verification-ladder.md):

• content-multiset-identity gate (file-split/extraction under @layer) — when a refactor moves rules between partials (not just reorders one), the byte-identical compile gate FAILS: each new partial adds a legitimate @layer base { } wrapper. The correct proof: strip @layer …{ openers and lone } lines, sort, diff → identical content-multiset proves no rule changed/added/dropped while the wrapper count differs as expected. Proven 5× (the _utilities dissolution). PROPOSED.
• CAUTION: property-name-exact collision measures miss shorthand↔longhand + cross-selector-same-specificity — the .ornament margin flip (margin shorthand vs margin-top/bottom longhand) was invisible to the automated /tmp/collide.py; only the source-read caught it. The automated census is the floor; the per-move render gate is the guarantor (byte-identical≠rendered-identical, one layer down). PROPOSED.

Memory-index (MEMORY.md) — the archive pass + the durable fixes:

• DONE 2026-06-08 (steward-authorized, live): compress-in-place archive pass. 90 archived-session bullets compressed to one-line pointers (detail preserved in the session files), 285KB→157KB, all 239 pointers preserved, every load-bearing section byte-identical; backup /tmp/MEMORY.md.bak. ⚠ A naive split-at-first-## Archived was caught-and-avoided — the file is INTERLEAVED (archived blocks scattered among load-bearing reference sections: Pending work / Feedback / Drift patterns / ARC project state / CapableMind L1 / Personal Context / Reference Files). Surgical compression of only archived bullets is the correct tool for an interleaved index.
• [PATCH proposal] /wrap-up §3 — compress at demotion. When demoting Active Session → Archived, compress the entry to a one-liner in the same move (it's already in the session file). Prevents index re-bloat at the source — without it the index re-grows ~4–5KB/session and this pass recurs.
• [PROPOSAL] deeper MEMORY.md reduction (only if 157KB still trips "only part loaded"). The load-bearing REFERENCE sections (steward profile / project states / ARC project state / CapableMind L1 / Reference Files) are large but largely STABLE — candidate to move to a consult-on-demand MEMORY-reference.md, leaving the wake-loaded MEMORY.md = standing prefs + trackers + Active Session + recent. Riskier (real memory, not duplication); its own focused pass, not a tail-of-session move.
• [reminder] the wake-canary (PROPOSED 2026-06-07, above) still awaits authorization — the paired guardrail (every MEMORY.md pointer + [[link]] resolves at wake); should cover MEMORY.md + any future archive/reference split files.

New proposals (2026-06-09 — the headless-Chrome measurement harness, awaiting steward)

Element	Kind	One-line	Where it lands	Status
Headless-Chrome box-model measurement	create skill OR verification-ladder note	When a rendered layout differs and the cause isn't obvious, measure the box model before theorizing the mechanism. Today this caught the phone-compass cause (per-point padding asymmetry, not width) after causal-story-before-reading-render recurred 2× (a footer-only first pass, then a wrong align-self:stretch width-fix). Reusable core: npm i puppeteer-core in a tmp dir · executablePath = /Applications/Google Chrome.app/... · setViewport({width:375,…,isMobile:true}) · page.evaluate reading getComputedStyle/getBoundingClientRect (widths, padding, row counts). Chrome reproduced the bug → it wasn't the Safari quirk I'd hypothesized.	a /measure-render skill, or a verification-ladder entry ("measure the box model before theorizing a layout difference")	PROPOSED
Symmetria §3 flag: theorize-before-measuring-a-layout	Symmetria §3 flag	The drift this caught, as a standing flag: a causal story about why a layout renders as it does, asserted before the rendered box model is measured, is contamination shape (same family as causal-story-before-reading-render, one step more specific — layout/CSS). Antidote: the harness above.	Symmetria §3	PROPOSED (may be redundant with the existing causal-story-before-reading-render flag — steward to judge)

New proposals (2026-06-09 pm — the typographic audit / measure / soft-rag day)

Element	Kind	One-line	Where it lands	Status
/measure-render (headless-Chrome box-model harness)	create skill OR ladder note	RE-FLAG — strongly earned. Proposed this morning (compass fix); the pm session ran it 4+ more times (true-advance measurement, viewport resize sweep, auto-vs-none hyphenation test). Proven repeatedly across one day. Caveat learned: headless Chrome cannot test hyphenation — it ships no hyphenation dictionaries (all configs gave identical heights); the real browser is the only valid hyphenation test.	/measure-render skill OR verification-ladder	PROPOSED (reinforced)
Measure the font's true average prose advance before a character-count measure	verification-ladder entry	When setting a measure to a target character count, measure the font's average advance over real corpus prose (incl. spaces), not the 0.5em convention. The convention misestimated EB Garamond by 32% (0.377em measured vs 0.5em assumed → the column ran ~90 chars, not the believed ~68). The instrument that overturned ARC's measure.	reference-verification-ladder.md	PROPOSED
ARC build has NO autoprefixer — hand-write -webkit- prefixes	feedback memory	ARC's plain-sass build adds no vendor prefixes. When introducing a new CSS property, check Safari's prefix need and hand-write it. The -webkit-hyphens: auto Safari bug came from prefixing the limits but not the property — an inconsistency that read as "hyphenation off in Safari."	feedback-arc-no-autoprefixer-handwrite-webkit.md	PROPOSED
Justification-judgment bar = Bringhurst even-colour/rivers, NOT Rutter hyphenation	feedback memory	Load-bearing for the future justification decision: when living with the soft rag to judge whether to justify, ask "is the colour even, are there rivers" (Bringhurst paragraph-level), not "is it hyphenating" (Rutter). The jurist corrected the executor's Rutter-leaning rationale. Either earns the KP/JS gate-amendment or retires it.	feedback-justification-bar-bringhurst-not-rutter.md	PROPOSED

New proposals (2026-06-10 — Stage-G close + spec-internal consistency audit day)

Element	Kind	One-line	Where it lands	Status
spec↔spec coherence dimension	patch /spec-code-audit	The 2026-06-10 audit found §I.f-class contradictions — the measure (38→27.2rem) un-propagated across silence-and-rhythm/apparatus/vignette, an American-quote remnant, v0.5/AUTHORED status leftovers — that bidirectional spec↔code could not catch, and that the jurist's read-scope (charter+multilingual only) missed. Add a cross-document spec↔spec coherence pass: every cross-referenced rule agrees across files; every late decision swept across ALL companions, not just its primary home. Proven method: a per-file-vs-canon + cross-reference fan-out (6 read-only agents).	/spec-code-audit (new dimension)	PROPOSED
container-must-embody-the-contained	feedback memory	When producing an ARTIFACT of a spec (a PDF of the spec, a rendered sample), set it per the spec's OWN rules and verify the render against the doctrine, not for approval. 2026-06-10: the Codex PDF violated its own §I.d (bold), §I.i (justified), §I.b (lining figures), §I.h (marked blockquotes), §VII.e (grey), and earlier the whole-document small-caps — and I read each render favourably instead of testing it against the spec (the SC miss especially). Sibling of causal-story-before-reading-render / read-render-favorably, one level up: the artifact must obey what it states.	feedback-container-must-embody-the-contained.md	PROPOSED
check-for-governed-tooling-before-building	feedback memory OR Symmetria §3 flag	Before hand-rolling infrastructure (a PDF preamble, a build script, a template), grep the repo for an existing governed version. 2026-06-10: I reinvented a lualatex preamble while tools/latex/arc-typography.sty (the governed PDF/print substrate, encoding the very rules I was hand-fixing) existed; the steward's "did we not do a basic pdf spec?" caught it. Kin to reinvented-governed-tooling-without-checking-it-exists (KG drift, 2026-06-10).	feedback memory or Symmetria §3	PROPOSED
byte-identical gate: hold/exclude volatile build-stamps	verification-ladder refinement	When proving a change byte-identical, a build-date/commit stamp (e.g. the colophon _build_info/stamp) will differ between baseline and post builds and falsely flag a diff. Hold the stamp constant (don't re-run make build-info) or exclude the stamped file, then diff. Caught 2026-06-10 (G1 proof: the lone "diff" was the colophon stamp).	reference-verification-ladder.md	PROPOSED

Not a skill — a parked project task (in the ledger + session file): the Phase-2 print/PDF-spec session to update arc-typography.sty + template-pdf-posture.tex to current doctrine (posture-aware ragged/justified, Medium headings, 27.2rem measure-geometry, British quote-normalization, xelatex glyph routing, v0.3). Governs all future ARC PDFs.

---

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?

New proposals (2026-06-11 — §VII.f / open-work-reckoning day, awaiting steward)

Element	Kind	One-line	Where it lands	Status
Symmetria §3 flag: inherited-marker-read-as-current-state	Symmetria §3 flag	A status/marker inherited from a RECORD (a selector-index entry, a tracker line, a "-pending" file, a prior framing) asserted as current state without verifying against code/live is contamination shape — records drift, in BOTH directions (stale "pending" files describe DONE work; "done" narratives hide OPEN work). Earned hard 2026-06-11: 3 stale-index misreads in one turn (running-head index entry, the "only unbuilt element" overclaim, the colophon) + the steward's two corrective catches (colophon done-mislabeled-pending; cul-de-lampe open-mislabeled-done). Antidote: verify status against the substrate before asserting. Kin to live-state-discipline / reasoning-from-remembered-model-not-current-state, generalised to ALL records.	Symmetria §3	PROPOSED
/measure-render (headless-Chrome box-model/scroll harness)	create skill OR ladder note	RE-REINFORCED (4th day of evidence). Proposed 2026-06-09 (compass fix) + reinforced 06-09 pm; today drove the ENTIRE §VII.f build — reveal opacity across scroll, body-column-axis alignment (598/798 constant), the dissolve-below-rule geometry, the frontispiece no-scroll proof, the compass graded reveal. The build would not have been groundable without it. Earns its own skill.	/measure-render skill OR verification-ladder	PROPOSED (strongly reinforced, 4th instance)

Watch-list (not yet a skill proposal): the open-work register pattern — when tracking surfaces drift, consolidate them into ONE code-verified source of truth (built project-arc-open-work-register.md today). Overlaps /spec-code-audit for the spec↔code half; the new part is the content/governance/Phase-2 reconciliation. May graduate to a /reconcile-open-work discipline if it recurs on another workstream (L1/be).