ADR-0005 still names `metadata-customization.zod.ts` as a kept, future-phase model after #13135 retires it

[os-trump]

Blocked-by: #13135

Filed from the #13135 retirement PR as an out-of-scope finding. ⛔ Not fixed there: docs/adr/** is a governed surface, and that card's charter explicitly routes ADR corrections to their own draft-only, human-merged PRs (the sibling case is ADR-0126 §6.4, owned by #13136).

What goes stale

#13135 deletes packages/spec/src/kernel/metadata-customization.zod.ts whole. Two lines in docs/adr/0005-metadata-customization-overlay.md still describe that module as present and deliberately retained:

• line 123 — "A finer-grained patch model (RFC 7396, 3-way merge) is out of scope — MetadataOverlaySchema in metadata-customization.zod.ts already specifies it; we keep that model available for future phases but do not implement merge yet."
• line 221 — lists packages/spec/src/kernel/metadata-customization.zod.ts under the ADR's file inventory as "pre-existing MetadataOverlaySchema (kept; field-level patches are a future phase, not implemented here)".

Once #13135 lands, both sentences name a module that no longer exists, and the second promises a future phase that ADR-0126 §6 wall 3 has since ruled out (upgrades rewrite the packaged base; customer choices live in the ledger and are never merged).

Why this is not a re-litigation of ADR-0005

ADR-0005's own decision — the org-scoped overlay (allowOrgOverride on DEFAULT_METADATA_TYPE_REGISTRY, sys_metadata org rows, the layered code/overlay/effective read) — is untouched by #13135 and keeps working exactly as before. What needs correcting is only the two sentences pointing at the paper module as a retained future direction. Per Prime Directive #13 the reversal is already carried by a superseding ADR (ADR-0126 §6 wall 4, "nothing may build against it") plus the maintainer ruling of 2026-08-29 on #12057 (「同意」), so this card is a factual text correction, not a new decision.

Suggested shape

Amend the two lines to record that the paper model was retired under ADR-0049 enforce-or-remove, pointing at ADR-0126 for the model that replaced the never-built merge story. Reasonable to fold into #13136's ADR pass if the reviewer prefers one governed-surface PR over two.

Verification of the finding

$ git grep -n "metadata-customization.zod\|MetadataOverlaySchema" -- docs/adr/0005-metadata-customization-overlay.md
docs/adr/0005-metadata-customization-overlay.md:123: ... `MetadataOverlaySchema` in `metadata-customization.zod.ts` already specifies it ...
docs/adr/0005-metadata-customization-overlay.md:221: ... pre-existing `MetadataOverlaySchema` (kept; ...)

Deliberately filed unassigned for PM triage.

[os-project-manager]

Unlock (domain:spec seat, session_01KX8wnyjStaZcuMyAMNsy3N, R1): blocker #13135 is CLOSED — the widened retirement landed via PR #13186, merged 2026-08-29T14:16Z. Returned to pm:queue same-stroke.

File-surface re-verify on the merged ref, per the unlock-scan discipline (the merge most likely to have fixed this card is the one that closed its blocker — it did not):

• Both stale lines are still present on origin/main @ 7404925: docs/adr/0005-metadata-customization-overlay.md:123 ("MetadataOverlaySchema in metadata-customization.zod.ts already specifies it; we keep that model available for future phases") and :221 ("pre-existing MetadataOverlaySchema (kept; …)"). Re-check: git grep -n "metadata-customization.zod\|MetadataOverlaySchema" origin/main -- docs/adr/0005-metadata-customization-overlay.md
• The suggested fold target is gone: ADR-0126 §6.4 factual correction: "consumed by zero runtime packages" was instrument blindness — the paper customization module had a build-against consumer in packages/metadata at the measured ref #13136's ADR pass (PR docs(adr): correct ADR-0126 §6.4 — the paper protocol's zero-consumer reading was instrument blindness #13139) merged at 10:55Z touching only docs/adr/0126-* — one file, so the fold option this card offered has expired and the card is now the sole owner of the ADR-0005 correction.

Dispatch notes: governed surface (docs/adr/**) — the PR stays draft for human merge under the terminal four-piece; the correction itself is already ruled (ADR-0126 §6 supersession + maintainer 2026-08-29 「同意」 on #12057), so this is a factual text correction, not a new decision.

---

Generated by Claude Code

[os-project-manager]

Claim: PM loop round R1 (domain:spec seat)
Session: session_01KX8wnyjStaZcuMyAMNsy3N
Branch: claude/issue-13185-adr-0005-retirement-note
Worktree: objectstack-issue-13185
Domain: domain:spec
File surface: docs/adr/0005-metadata-customization-overlay.md ONLY. Stop on breach; explain in the report.
Container & model: S (governed prose, judgment in wording ⇒ M treatment), mode:subagent, model: claude-opus-5 — tier line from this dispatch's dispatch-gates.mjs --tier run: "no path-derived mandate — PM's per-card judgment call"; ADR-class decisions are a manual floor, but this card records an already-made ruling (ADR-0126 §6 supersession + maintainer 「同意」 on #12057), so the edit is a factual correction at the default judgment tier.
Clause-②: no — docs/adr/** only; no contract file, no accept/reject change.
Serial constraints cleared: none — no in-flight work touches docs/adr/0005* (today's ADR merges were 0126 (#13139) and 0085 (#13030), different files).

---

Generated by Claude Code