Draft: Memento for Machines (model-authored essay — HOLD for captain review)

[git-repo-auth]

HOLD — captain review required (gauntlet step 7 is constitutionally his).

Model-authored essay rolling the Memento + Alzheimer's framings into the augmented-memory architecture. Publish-gauntlet evidence trail (run as remediation — drafting preceded the law-fetch, itself now a debrief candidate):

1. Preflight before drafting -> MISSED, disclosed: gauntlet fired post-draft; per canon this run is remediation, recorded as such.
2. Law fetched live: writing-canon (hash 6bhvt7), publish-gauntlet (wnugeq), triangle-pass, guide-posture, dual-context-writing, ai-voice-cliches — all in-session, at canon sha 99192cd.
3. Structure to standard: stance title; Tier-2 blockquote rewritten as a complete compressed argument; ## Summary — extraction key added, opening on the reader's pain (guide posture); headers tell the story in sequence.
4. Mechanical audit (numbers): 37 em-dashes / 43 paragraphs, 2 paragraphs at 3+; negation-parallelism ~2; formulaic-transition/cliche hits 0; sentence pacing mean 19.1w stdev 14.6.
5. Challenge, published-essay mode: 7 claim-types matched, 27 questions; 3 material findings addressed in-text (comparative fairness + as-of date; extended-mind prior art; explicit retraction condition). Remainder declined as genre-inappropriate hedge, per the gauntlet's calibration clause.
6. Disclosure by extraction: blockquote and Summary tiers verified present and independently actionable by direct read; live read-model extraction is only possible post-merge (pre-merge limitation, disclosed).
7. Author's pass: PENDING — this hold.
8. This PR body is the trail.

Receipts-appendix visibility claims re-verified unauthenticated (trust-kernel 200, ODD candidates 200, ARS 404-as-expected). Frontmatter validator: green, including the custom receipts block.

---

Note

Low Risk
Content-only addition under writings/ with no application or auth changes; main risk is editorial/publish posture and accuracy of attested claims, not production behavior.

Overview
Adds writings/memento-for-machines.md, a new Tier-2, public, first-person essay that argues AI sessions are structural anterograde amnesia and maps an external hippocampus architecture (debrief → candidate queue → distillation → versioned canon → mandatory boarding recall), guarded by fresh-context validation, two-party promotion, and content-addressed serving.

The piece uses Memento’s four infrastructure failures and a caregiving/Alzheimer’s framing to explain why unaudited “memory features” fail, then ties claims to canon URIs and a receipts frontmatter ledger (public and attested private refs). It includes a retraction condition for native model memory and ends with an explicit Receipts appendix; merge is blocked on captain review per the publish gauntlet (author’s pass pending).

^{Reviewed by Cursor Bugbot for commit 1c025e6. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

Canon Quality — Frontmatter Schema ✅

All 50 file(s) in writings/ conform to klappy://canon/meta/frontmatter-schema.

_{Validator: scripts/validate-frontmatter.py · Canon: klappy://canon/constraints/frontmatter-validation-before-merge · Run: #392}

[github-actions]

Canon Quality — P0010 Retrieval-Readiness ⚠️

Soft report for klappy://canon/constraints/retrieval-disclosure-contract. 702 files scanned. Never blocks — informational until the corpus is ready to enforce.

• Blocking-class findings: 15 (structural fields the contract would filter on)
• Warnings: 0 (kind resolves to unknown)
• Informational: 13 (exempt templates/archive/drafts)

Kind distribution: {'essays': 52, 'canon': 242, 'apocrypha': 38, 'docs': 304, 'journals': 60, 'unknown': 6}
Kind source: {'path': 567, 'frontmatter': 129, 'none': 6} (frontmatter-primary, path-secondary)
Default-include visibility: 598 visible, 104 hidden (journals/apocrypha/unknown)

By rule: {'audience-invalid': 2, 'exposure-missing': 5, 'tier-missing': 5, 'tier-invalid': 7, 'fm-missing': 3, 'kind-unresolvable': 6}

These are not schema violations (see the Frontmatter Schema job for those on writings/). They are corpus-readiness signals for the retrieval contract: invalid/missing audience, exposure, tier, and docs whose kind cannot be resolved. Fix in a corpus-cleanup PR before the contract flips to enforcing. See the retrieval-readiness-findings artifact for the full list.

_{Validator: scripts/audit-retrieval-readiness.py · Constraint: klappy://canon/constraints/retrieval-disclosure-contract · Run: #392}

[github-actions]

Canon Quality — Homepage Surfacing ✅

50 essay(s) scanned. Soft report — never blocks; the hard field gate is the Frontmatter Schema job.

All published essays resolve to the homepage feed.

_{Report: scripts/surfacing-report.py · Canon: klappy://canon/constraints/frontmatter-validation-before-merge}

[github-actions]

Canon Quality — oddkit_audit ✅

No dead klappy:// references or legacy link patterns found in writings/. 52 files scanned.

_{Spec: klappy://docs/oddkit/specs/oddkit-audit · Workflow: .github/workflows/canon-quality.yml · Run: #392}