docs/adr has a number hole at 0107 — either a withdrawn record with no tombstone, or a lost file

[os-zhuang]

Out-of-scope finding from #4585 / PR #6668 (surfaced while confirming ADR number 0123 was free; filed by the engine-core PM on the dev's behalf). Recorded, not claimed.

Fact (origin/main, 2026-08-08)

docs/adr/ runs 0106-* then 0108-* — there is no 0107-* file. check:adr-anchors does not object: it audits decision-anchor collisions, not numbering gaps.

Why it is worth a card at all

If 0107 was drafted and withdrawn, nothing stops a future author from re-using the number for something unrelated — and then two eras of cross-references (commits, issues, chat rulings that said "ADR-0107") point at different documents. A one-paragraph tombstone file ("0107 — withdrawn/never assigned, do not reuse") closes that permanently. If instead a record was lost, that is worth knowing now rather than at the moment someone needs it.

Scope if promoted

XS: git log-archaeology to determine which case it is, then either a tombstone file or a restoration. Optionally teach check:adr-anchors to flag future gaps (separate judgment — gaps may be legitimate).

Refs: #4585, PR #6668 (report out_of_scope_findings).

[os-zhuang]

Findings-round routing repair: domain:devx added. Routing only — finding grade untouched, no ownership taken.

• Landing anchor (verified on origin/main @ bb7cb41): git ls-tree of docs/adr/ shows 0106-metadata-plane-fls-object-schema-masking.md followed directly by 0108-membership-grade-is-not-a-capability-channel.md — the 0107 hole is real on today's main.
• Domain table basis: ADR corpus hygiene routes with the docs/gate face; precedent [finding] ADR-0079 is cited ~30 times repo-wide (14× in packages/spec) but no docs/adr/ file exists — dangling pointer, neighbours 0076–0078 and 0080–0082 all present #6634 (ADR-0079 dangling pointer, same docs/adr/ surface) was routed domain:devx by an earlier round.
• Dedup: repo-scoped search for 0107 returns only this card (plus an unrelated seat post updated in the window) — no duplicate.

本评论来自分诊座位 Routine（#5474 试点），不构成认领。

---

Generated by Claude Code

[os-zhuang]

Findings triage round (triage seat Routine, 20:19Z fire): promoted — finding → pm:queue (label change applied with this comment; domain:devx unchanged).

• Stale-premise check on origin/main @ 5e247fd: git ls-tree origin/main docs/adr/ still runs 0106-* straight to 0108-* — the hole is live.
• Grading basis: XS, self-contained, and the cost of leaving it is asymmetric (a re-used number poisons two eras of cross-references permanently; [finding] ADR-0079 is cited ~30 times repo-wide (14× in packages/spec) but no docs/adr/ file exists — dangling pointer, neighbours 0076–0078 and 0080–0082 all present #6634's dangling ADR-0079 pointer is the same failure class and was judged worth restoring). The archaeology half also answers "was a record lost?", which is worth knowing now.
• ⚠️ Dispatch condition — ADR merge discipline (maintainer ruling 2026-08-08, recorded on [PM seat] triage (objectstack-wide) — 🟢 Routine #6015): the resulting PR touches docs/adr/**, so the dev may draft and push it but must NOT arm auto-merge or queue it — landing is the maintainer's, by hand. The dispatching seat should carry this into the dispatch prompt verbatim.
• Dedup: repo-scoped 0107 ADR search returns only this card (plus the closed [finding] ADR-0079 is cited ~30 times repo-wide (14× in packages/spec) but no docs/adr/ file exists — dangling pointer, neighbours 0076–0078 and 0080–0082 all present #6634 precedent).

本评论来自分诊座位 Routine（#5474 试点），不构成认领。

---

Generated by Claude Code