Every long-form gate agent sees ONE edition, so «the reader has heard this
before» was structurally invisible — and the specifics-bank dedupe actively
encouraged re-surfacing the same material. At series cadence that is the
fastest-growing defect class. N11 makes it visible BEFORE the skeleton is
approved, at two grains:
- Series grain — new scripts/editions package: each locked edition's spent
anecdotes/arguments/hooks are written to <serie>/linkedin/series-distillate.json
at Step 8 lock (distil-append), and the next skeleton is checked against it at
Step 2.5 (distil-check) with the finding folded into the annotation gate the
operator already reads. Advisory, never blocking: a deliberate callback is a
legitimate move, an unnoticed retread is not.
- Material grain — specifics-bank usedIn log: record-usage stamps «used in
edition NN» on the specifics an edition actually consumed (read from the bound
slot-map, so abstrakt/ekstern stamp nothing). At lock, so it means published;
idempotent under a pivot re-lock. Additive-optional, schema stays v1.
Placement deviates from the plan text (${DATA}) after premise-verification:
per-series state belongs in the series root beside edition-state.json, where
Step 0 already resolves the path and no slug→path map is needed. Operator
approved. The distillate module also lands in scripts/editions now rather than
at N12, since the AC required a testable roundtrip and N12 planned that package
anyway — N12 extends it instead of creating it.
Similarity is character-trigram Jaccard, not word overlap: the plugin is
language-general and inflection (migrere/migreringen) breaks word tokens.
Calibrated on real paraphrase pairs — retellings 0.44-0.55, unrelated 0.04-0.06,
same-topic-different-story 0.24 — so the default threshold sits in the gap at
0.40. Word-Jaccard scored a shortened hook paraphrase at 0.11.
TDD (Iron Law): tests written first and verified red before implementation, for
both the new package and the bank logging.
Suites: editions 27/0 (new suite line + guard, floor 27) · specifics-bank
28 -> 45 · test-runner 163 -> 173 (Section 16r: 9 unconditional greps +
non-vacuity self-test; anti-erosion floor 146 -> 155) · trends 300/0 ·
brain 134/0 · hooks 140/0 · tests 35/0 · render 60/0. tsc --noEmit clean.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QxvWAjte7vPcF79QeSRvRJ
Slice 2 of fix#2 (kilde-så-draft): the per-edition bridge between the global
specifics-bank (slice 1) and a draft. In scripts/specifics-bank:
- binding.ts — livedSpecifics slot-map model + validateBinding, the «vaghet
avvises» gate (G4). BLOCK on unresolved / dangling specific-id / unjustified
abstrakt; WARN on an unverified-number-backed slot (regel 6/7); PASS otherwise.
- kilder.ts — renderKilder, the read-only NN-kilder.md sources ledger, rendered
deterministically from the slot-map + bank (coverage via validateBinding, so
the artifact and the skeleton-gate cannot disagree).
- cli.ts — validate-binding (exit 1 = BLOCK) + render-kilder subcommands.
edition-state schema gains articles.NN.livedSpecifics ({ slots, status }) + _doc.
Design record + the Step 2 research re-scope spec: docs/fix2/slice2-binding.md.
Wiring into newsletter.md (Step 1.5 elicitation + Step 2.5 gate + pipeline 17->18)
is slice 3 — not in this commit.
Tests 28/28 (specifics-bank; +9 binding +4 kilder), tsc clean, plugin gate 83/0/0
unaffected, gitleaks clean on all 8 files.
[skip-docs] internal plumbing — no command/agent/pipeline surface change until
slice 3 wires it; documenting it in the user-facing root CLAUDE.md/README now would
overclaim unreachable functionality. Docs ARE updated where they belong: the design
record docs/fix2/slice2-binding.md, scripts/specifics-bank/README.md, and the
edition-state _doc. Root-doc update lands with the slice-3 wiring feat.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Fix#2 (kilde-så-draft) slice 1: the deterministic STORE half of the
lived-specifics bank — the operator's real, un-generatable raw material
(measured number / named-but-real case / what broke / contrarian opinion /
mind-change), so drafts start grounded instead of hollow (retning §3;
dream-spec "Lived-Specifics Extraction", the single biggest missing upstream
piece for the authenticity thesis).
scripts/specifics-bank/ (sibling to contract-gate, same tsx convention):
- src/types.ts — Specific/Bank schema (schemaVersion 1), un-generatable taxonomy
- src/bank.ts — pure store: normalizeContent, content-hash id (= dedupe key),
load/save, addSpecific (dedupe + topicTag union on re-capture), queryByTopic
(active-only, ranked by tag overlap then recency)
- src/cli.ts — query / add / list; default bank under
${LINKEDIN_STUDIO_DATA:-~/.claude/linkedin-studio}/specifics-bank/ so it
survives plugin upgrades/reinstalls (M0 data-path seam)
- tests/bank.test.ts — 15/15 green; tsc clean
- README + .gitignore for node_modules/build
Content is stored verbatim; never AI-generated. The elicitation interview
(refuses vague answers; abstrakt/ekstern escape hatches) lands in slice 3
(/linkedin:newsletter Step 1.5). Forward-compatible with the profile-evolution
/ "second brain" architecture.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
