<?xml version="1.0" encoding="utf-8"?>
<entry xmlns="http://www.w3.org/2005/Atom" xml:base="https://docs.spec-kitty.ai/adr/3.x/2026-08-03-1-review-cycle-artifacts-are-coord-partition/">
  <id>https://docs.spec-kitty.ai/adr/3.x/2026-08-03-1-review-cycle-artifacts-are-coord-partition/</id>
  <title>ADR: Review-Cycle Artifacts Are COORD-Partition Per-WP Bookkeeping</title>
  <updated>2026-08-03T00:00:00.000Z</updated>
  <author><name>Spec Kitty</name></author>
  <link rel="alternate" type="text/html" href="https://docs.spec-kitty.ai/adr/3.x/2026-08-03-1-review-cycle-artifacts-are-coord-partition/"/>
  <summary>Review-cycle artifacts get a `REVIEW_CYCLE` kind in the COORD partition — per-work-package lifecycle evidence, not stable planning output — visible to cross-lane reviewers.</summary>
  <content type="html">&lt;p&gt;&lt;strong&gt;Deciders&lt;/strong&gt;: operator (stijn-dejongh), overriding a squad CRITICAL that had recorded
the partition as an unresolved contradiction. The adjudication, verbatim: &lt;em&gt;“Land them
on COORD in case of a coord topology (this is per-wp bookkeeping, and should be
traced in the shared Coord branch, which is aggregated into the primary branch on
consolidation). In other topologies -&gt; primary. Ensure read and write paths are
consistent.”&lt;/em&gt;&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Revision note&lt;/strong&gt;: a first draft of this ADR was rejected by a two-lens adversarial
check. Its factual errors are recorded in “What the first draft got wrong” below
rather than silently corrected, because two of them would have produced an
unimplementable change.&lt;/p&gt;
&lt;h2 id=&quot;context&quot;&gt;Context&lt;/h2&gt;
&lt;p&gt;Two authority sources answered the same question in contradictory ways, and the
contradiction was load-bearing rather than cosmetic.&lt;/p&gt;
&lt;p&gt;The canonical glossary — &lt;a href=&quot;/context/orchestration/#coord-partition&quot;&gt;&lt;code dir=&quot;auto&quot;&gt;docs/context/orchestration.md#coord-partition&lt;/code&gt;&lt;/a&gt;,
an authority path named in the project charter — defines the COORD partition as
holding &lt;em&gt;“a mission’s lifecycle/coordination artifacts — status events, notes,
trace, issue-matrix, acceptance-matrix, &lt;strong&gt;review cycles&lt;/strong&gt;, &lt;code dir=&quot;auto&quot;&gt;move-task&lt;/code&gt;”&lt;/em&gt;.&lt;/p&gt;
&lt;p&gt;The code says otherwise. There is &lt;strong&gt;no &lt;code dir=&quot;auto&quot;&gt;REVIEW_CYCLE&lt;/code&gt; artifact kind&lt;/strong&gt;
(&lt;code dir=&quot;auto&quot;&gt;src/mission_runtime/artifacts.py&lt;/code&gt;). Review-cycle artifacts borrow
&lt;code dir=&quot;auto&quot;&gt;MissionArtifactKind.WORK_PACKAGE_TASK&lt;/code&gt;, a member of &lt;code dir=&quot;auto&quot;&gt;_PRIMARY_ARTIFACT_KINDS&lt;/code&gt;, and
&lt;code dir=&quot;auto&quot;&gt;src/specify_cli/post_merge/review_artifact_consistency.py&lt;/code&gt; asserts in its own
docstring that they are &lt;em&gt;“&lt;code dir=&quot;auto&quot;&gt;WORK_PACKAGE_TASK&lt;/code&gt;, PRIMARY-partition for &lt;strong&gt;every&lt;/strong&gt;
topology”&lt;/em&gt;.&lt;/p&gt;
&lt;p&gt;The glossary was also &lt;strong&gt;not unanimous with itself&lt;/strong&gt;: its PRIMARY entry, describing
the same COORD set, omitted review cycles and acceptance-matrix. That entry is
corrected in the same change as this ADR, so the two now agree.&lt;/p&gt;
&lt;h3 id=&quot;why-this-became-load-bearing&quot;&gt;Why this became load-bearing&lt;/h3&gt;
&lt;p&gt;Mission &lt;code dir=&quot;auto&quot;&gt;review-cycle-verdict-seam-rebuild-01KZ2W7W&lt;/code&gt; makes the status &lt;strong&gt;event&lt;/strong&gt;
authoritative for &lt;em&gt;which&lt;/em&gt; verdict is current while the &lt;strong&gt;artifact&lt;/strong&gt; holds &lt;em&gt;what the
reviewer said&lt;/em&gt;. Under coord topology the event commits to the coordination branch
and the artifact — as &lt;code dir=&quot;auto&quot;&gt;WORK_PACKAGE_TASK&lt;/code&gt; — commits to the primary target branch, so
the two halves of one fact live on two branches. Two concerns in that mission’s plan
were found specifying &lt;strong&gt;opposite partitions for the same fact&lt;/strong&gt;.&lt;/p&gt;
&lt;h2 id=&quot;decision&quot;&gt;Decision&lt;/h2&gt;
&lt;p&gt;&lt;strong&gt;Review-cycle artifacts are COORD-partition per-work-package bookkeeping.&lt;/strong&gt; They are
lifecycle evidence about a work package’s passage through review, not stable planning
output: written repeatedly during execution, per-WP, and belonging on the shared coord
surface where a reviewer working across lanes can see them.&lt;/p&gt;
&lt;ol&gt;
&lt;li&gt;Introduce &lt;code dir=&quot;auto&quot;&gt;MissionArtifactKind.REVIEW_CYCLE&lt;/code&gt;. Review-cycle artifacts stop borrowing
&lt;code dir=&quot;auto&quot;&gt;WORK_PACKAGE_TASK&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;Place &lt;code dir=&quot;auto&quot;&gt;REVIEW_CYCLE&lt;/code&gt; in &lt;code dir=&quot;auto&quot;&gt;_PLACEMENT_ARTIFACT_KINDS&lt;/code&gt; (the COORD set).&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;The seam projection needs no change; the path classifier does.&lt;/strong&gt; Verified by
probe: &lt;code dir=&quot;auto&quot;&gt;resolve_artifact_surface&lt;/code&gt; already returns PRIMARY for a COORD kind under
&lt;code dir=&quot;auto&quot;&gt;SINGLE_BRANCH&lt;/code&gt; / &lt;code dir=&quot;auto&quot;&gt;LANES&lt;/code&gt;, so the topology rule falls out of set membership at the
&lt;em&gt;read&lt;/em&gt; seam. But the &lt;strong&gt;write&lt;/strong&gt; side does not follow, and this is a named
deliverable, not an implementation detail — see “Required machinery” below.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Read and write paths resolve the same kind.&lt;/strong&gt; Every producer and consumer
resolves through one owner function; a caller-supplied directory is not a
substitute. Nine of twelve current sites hard-code a PRIMARY assumption and must
be migrated together — a half-migrated seam is worse than either endpoint.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Conflict rule.&lt;/strong&gt; When both surfaces hold a record for the same work package, the
COORD copy is authoritative under a coordination topology. This inverts
&lt;code dir=&quot;auto&quot;&gt;test_review_artifact_gate_ignores_stray_artifact_on_coord_husk&lt;/code&gt; (harvested from
PR #2834), which pins the opposite. That test encodes the &lt;em&gt;pre-ADR&lt;/em&gt; partition and
is re-pinned citing this ADR, with a both-present case added. The old polarity was
correct while review cycles were PRIMARY; it is wrong once they are not.&lt;/li&gt;
&lt;/ol&gt;
&lt;p&gt;P-1 is preserved: &lt;code dir=&quot;auto&quot;&gt;REVIEW_CYCLE&lt;/code&gt; joins exactly one set, and &lt;code dir=&quot;auto&quot;&gt;WORK_PACKAGE_TASK&lt;/code&gt; — the
WP task file itself — stays PRIMARY.&lt;/p&gt;
&lt;h3 id=&quot;required-machinery&quot;&gt;Required machinery&lt;/h3&gt;
&lt;p&gt;The first draft claimed “no new routing machinery”. That was false. Three mechanisms
must change, and each is an acceptance item:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Path classifier.&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;_artifact_kind_for_path&lt;/code&gt; returns
&lt;code dir=&quot;auto&quot;&gt;_COORD_RESIDUE_DIRS.get(mission_rel_parts[0])&lt;/code&gt; for &lt;em&gt;anything&lt;/em&gt; nested under
&lt;code dir=&quot;auto&quot;&gt;tasks/&lt;/code&gt;, and its file leg is a fixed-basename dict. &lt;code dir=&quot;auto&quot;&gt;review-cycle-&amp;#x3C;N&gt;.md&lt;/code&gt; has no
fixed basename, so &lt;code dir=&quot;auto&quot;&gt;REVIEW_CYCLE&lt;/code&gt; &lt;strong&gt;cannot be expressed in the classifier today&lt;/strong&gt;.
It needs a nested-pattern leg. The rule must be &lt;strong&gt;filename-anchored&lt;/strong&gt;
(&lt;code dir=&quot;auto&quot;&gt;review-cycle-*.md&lt;/code&gt;), not directory-anchored: &lt;code dir=&quot;auto&quot;&gt;tasks/&amp;#x3C;wp&gt;/&lt;/code&gt; also holds
&lt;code dir=&quot;auto&quot;&gt;baseline-tests.json&lt;/code&gt;, which is deliberately PRIMARY, and &lt;code dir=&quot;auto&quot;&gt;tasks/WP*.md&lt;/code&gt; must stay
&lt;code dir=&quot;auto&quot;&gt;WORK_PACKAGE_TASK&lt;/code&gt;.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Commit router.&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;_group_files_by_partition&lt;/code&gt; re-classifies by path and &lt;strong&gt;overrides
the caller’s kind&lt;/strong&gt; — its docstring: &lt;em&gt;“The buckets are ABSOLUTE, not relative to the
caller.”&lt;/em&gt; Without the classifier leg, a &lt;code dir=&quot;auto&quot;&gt;kind=REVIEW_CYCLE&lt;/code&gt; write is re-bucketed to
PRIMARY, the file is absent from the primary ref, the result is
&lt;code dir=&quot;auto&quot;&gt;no_op_wrong_surface&lt;/code&gt;, and &lt;code dir=&quot;auto&quot;&gt;_commit_review_cycle_artifact&lt;/code&gt; escalates to a hard error
&lt;strong&gt;and unlinks the artifact&lt;/strong&gt;. Every rejection write would fail.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;E2 eligibility.&lt;/strong&gt; &lt;code dir=&quot;auto&quot;&gt;_E2_CONSOLIDATED_ELIGIBLE_KINDS&lt;/code&gt; currently covers review cycles
via &lt;code dir=&quot;auto&quot;&gt;WORK_PACKAGE_TASK&lt;/code&gt;. An explicit ruling for &lt;code dir=&quot;auto&quot;&gt;REVIEW_CYCLE&lt;/code&gt; is required, or a
PUBLISHED mission’s write falls through to an unconditional coordination probe.&lt;/li&gt;
&lt;/ul&gt;
&lt;h3 id=&quot;migration-exception-absorption-not-empty-directory-fallback&quot;&gt;Migration: exception absorption, not empty-directory fallback&lt;/h3&gt;
&lt;p&gt;The first draft said a read finding &lt;em&gt;nothing&lt;/em&gt; at COORD falls back to PRIMARY. That is
not the failure mode. Measured against this repository:&lt;/p&gt;
&lt;div class=&quot;expressive-code&quot;&gt;&lt;figure class=&quot;frame not-content&quot;&gt;&lt;figcaption class=&quot;header&quot;&gt;&lt;/figcaption&gt;&lt;pre data-language=&quot;plaintext&quot;&gt;&lt;code&gt;&lt;div class=&quot;ec-line&quot;&gt;&lt;div class=&quot;code&quot;&gt;&lt;span style=&quot;--0:#d6deeb;--1:#403f53&quot;&gt;missions w/ review cycles: 102        (418 artifacts)&lt;/span&gt;&lt;/div&gt;&lt;/div&gt;&lt;div class=&quot;ec-line&quot;&gt;&lt;div class=&quot;code&quot;&gt;&lt;span class=&quot;indent&quot;&gt;&lt;span style=&quot;--0:#d6deeb;--1:#403f53&quot;&gt;  &lt;/span&gt;&lt;/span&gt;&lt;span style=&quot;--0:#d6deeb;--1:#403f53&quot;&gt;declare a coordination_branch: 45&lt;/span&gt;&lt;/div&gt;&lt;/div&gt;&lt;div class=&quot;ec-line&quot;&gt;&lt;div class=&quot;code&quot;&gt;&lt;span class=&quot;indent&quot;&gt;&lt;span style=&quot;--0:#d6deeb;--1:#403f53&quot;&gt;    &lt;/span&gt;&lt;/span&gt;&lt;span style=&quot;--0:#d6deeb;--1:#403f53&quot;&gt;branch STILL EXISTS in git : 0&lt;/span&gt;&lt;/div&gt;&lt;/div&gt;&lt;div class=&quot;ec-line&quot;&gt;&lt;div class=&quot;code&quot;&gt;&lt;span class=&quot;indent&quot;&gt;&lt;span style=&quot;--0:#d6deeb;--1:#403f53&quot;&gt;    &lt;/span&gt;&lt;/span&gt;&lt;span style=&quot;--0:#d6deeb;--1:#403f53&quot;&gt;branch DELETED from git    : 45&lt;/span&gt;&lt;/div&gt;&lt;/div&gt;&lt;/code&gt;&lt;/pre&gt;&lt;/figure&gt;&lt;/div&gt;
&lt;p&gt;&lt;code dir=&quot;auto&quot;&gt;spec-kitty merge&lt;/code&gt; deletes the mission branch; the coordination branch &lt;strong&gt;is&lt;/strong&gt; the
mission branch; nothing clears the &lt;code dir=&quot;auto&quot;&gt;meta.json&lt;/code&gt; key. Every merged coord mission is
permanently in the seam’s &lt;code dir=&quot;auto&quot;&gt;DELETED&lt;/code&gt; cell, which is &lt;strong&gt;fail-loud by contract&lt;/strong&gt; and
pinned by &lt;code dir=&quot;auto&quot;&gt;test_deleted_coord_branch_raises_fail_loud&lt;/code&gt;. The seam raises
&lt;code dir=&quot;auto&quot;&gt;CoordinationBranchDeleted&lt;/code&gt; &lt;em&gt;before any read happens&lt;/em&gt;, and the merge gate calls it
unguarded.&lt;/p&gt;
&lt;p&gt;So the migration rule is: &lt;strong&gt;&lt;code dir=&quot;auto&quot;&gt;REVIEW_CYCLE&lt;/code&gt; reads absorb &lt;code dir=&quot;auto&quot;&gt;CoordinationBranchDeleted&lt;/code&gt;
and &lt;code dir=&quot;auto&quot;&gt;StatusReadPathNotFound&lt;/code&gt; to the PRIMARY directory&lt;/strong&gt;, in one owner function — not
per consumer, and not at the seam, which resolves a directory per kind and never
probes artifact existence. The house precedent is the issue-matrix JSON-first/&lt;code dir=&quot;auto&quot;&gt;.md&lt;/code&gt;
failover, which lives in the artifact’s owner module.&lt;/p&gt;
&lt;h2 id=&quot;consequences&quot;&gt;Consequences&lt;/h2&gt;
&lt;p&gt;&lt;strong&gt;Positive.&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Glossary and code agree, and the agreement is checkable because the kind exists.&lt;/li&gt;
&lt;li&gt;Verdict authority and verdict record land on the same partition under coord
topology, so the merge gate stops comparing across a branch boundary.&lt;/li&gt;
&lt;li&gt;The “opposite partitions for the same fact” contradiction dissolves.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;strong&gt;Negative, accepted.&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;45 existing coord missions&lt;/strong&gt; carry review cycles behind a deleted coordination
branch. They are served by exception absorption, permanently, unless a separate
&lt;code dir=&quot;auto&quot;&gt;meta.json&lt;/code&gt; flatten migration retires the stale key.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;&lt;code dir=&quot;auto&quot;&gt;tasks/&lt;/code&gt; becomes two-sided.&lt;/strong&gt; It is registered in
&lt;code dir=&quot;auto&quot;&gt;_NON_DIVERGENT_COORD_RESIDUE_DIRS&lt;/code&gt; on the justification that it is &lt;em&gt;“authored once
and never independently edited on the target side”&lt;/em&gt;. The flip falsifies that: WP
files on target, review cycles on the mission branch. During the migration window a
coord mission with cycles 1–3 on PRIMARY writes cycle 1 to COORD (&lt;code dir=&quot;auto&quot;&gt;next_cycle_number&lt;/code&gt;
globs one directory) and &lt;code dir=&quot;auto&quot;&gt;-X theirs&lt;/code&gt; overwrites the target’s cycle 1 — the #2804
clobber shape, in a directory the guard has pre-classified as safe. Either a
reconcile driver or a documented re-justification is required.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;The create window splits one WP’s artifacts.&lt;/strong&gt; The coord worktree materialises
lazily at the commit boundary, so a coord mission’s &lt;em&gt;first&lt;/em&gt; review cycle is written
to PRIMARY and later ones to COORD, with &lt;code dir=&quot;auto&quot;&gt;next_cycle_number&lt;/code&gt; counting only one
surface.&lt;/li&gt;
&lt;li&gt;One more kind, plus rows in &lt;code dir=&quot;auto&quot;&gt;PARTITION_RATIONALE&lt;/code&gt;
(&lt;code dir=&quot;auto&quot;&gt;tests/architectural/test_write_surface_placement_guard.py&lt;/code&gt;), which is pinned
exhaustive and will go red on the enum addition alone.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;&lt;strong&gt;Neutral.&lt;/strong&gt;&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;No change to &lt;code dir=&quot;auto&quot;&gt;WORK_PACKAGE_TASK&lt;/code&gt;’s placement, or for &lt;code dir=&quot;auto&quot;&gt;SINGLE_BRANCH&lt;/code&gt; / &lt;code dir=&quot;auto&quot;&gt;LANES&lt;/code&gt;
missions.&lt;/li&gt;
&lt;li&gt;Lane worktrees need &lt;strong&gt;no&lt;/strong&gt; sparse-checkout change. The first draft framed this
wrongly: a lane is checked out on the lane branch, so COORD content is absent
regardless of patterns. A lane reaches its review cycles through the seam, which
anchors on &lt;code dir=&quot;auto&quot;&gt;get_main_repo_root&lt;/code&gt; and is CWD-invariant.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id=&quot;what-the-first-draft-got-wrong&quot;&gt;What the first draft got wrong&lt;/h2&gt;
&lt;p&gt;Recorded because two of these would have shipped an unimplementable change.&lt;/p&gt;





























&lt;table&gt;&lt;thead&gt;&lt;tr&gt;&lt;th&gt;Claim&lt;/th&gt;&lt;th&gt;Status&lt;/th&gt;&lt;/tr&gt;&lt;/thead&gt;&lt;tbody&gt;&lt;tr&gt;&lt;td&gt;“No new routing machinery is required”&lt;/td&gt;&lt;td&gt;&lt;strong&gt;False.&lt;/strong&gt; True of the read seam only; the classifier and commit router both need work.&lt;/td&gt;&lt;/tr&gt;&lt;tr&gt;&lt;td&gt;“A read that finds nothing at COORD falls back to PRIMARY”&lt;/td&gt;&lt;td&gt;&lt;strong&gt;Not the failure mode.&lt;/strong&gt; The seam raises first, for all 45 affected missions.&lt;/td&gt;&lt;/tr&gt;&lt;tr&gt;&lt;td&gt;The &lt;code dir=&quot;auto&quot;&gt;slice-f&lt;/code&gt; on-disk evidence shows partition drift&lt;/td&gt;&lt;td&gt;&lt;strong&gt;Misattributed.&lt;/strong&gt; That mission has &lt;code dir=&quot;auto&quot;&gt;topology: None&lt;/code&gt; — no partition to disagree about. The real cause is &lt;code dir=&quot;auto&quot;&gt;arbiter.py&lt;/code&gt; probing bare &lt;code dir=&quot;auto&quot;&gt;wp_id&lt;/code&gt; while the writer uses &lt;code dir=&quot;auto&quot;&gt;wp_slug&lt;/code&gt;, already diagnosed under the mission’s IC-09. The evidence is struck.&lt;/td&gt;&lt;/tr&gt;&lt;tr&gt;&lt;td&gt;“The coord branch is aggregated into primary on consolidation”&lt;/td&gt;&lt;td&gt;&lt;strong&gt;True, but not by the assumed mechanism.&lt;/strong&gt; Not &lt;code dir=&quot;auto&quot;&gt;bookkeeping_projection.py&lt;/code&gt;, which is basename-whitelisted to two status files. The carry is the mission-branch squash: the coordination branch &lt;em&gt;is&lt;/em&gt; &lt;code dir=&quot;auto&quot;&gt;kitty/mission-&amp;#x3C;slug&gt;-&amp;#x3C;mid8&gt;&lt;/code&gt;, merged by &lt;code dir=&quot;auto&quot;&gt;_phase_mission_to_target&lt;/code&gt; via &lt;code dir=&quot;auto&quot;&gt;git merge --squash -X theirs&lt;/code&gt;. Verified on 94/96 real missions.&lt;/td&gt;&lt;/tr&gt;&lt;tr&gt;&lt;td&gt;The glossary “defines” review cycles as COORD&lt;/td&gt;&lt;td&gt;&lt;strong&gt;One-entry read&lt;/strong&gt; of a two-entry, self-disagreeing definition. Fixed in the same change.&lt;/td&gt;&lt;/tr&gt;&lt;/tbody&gt;&lt;/table&gt;
&lt;h2 id=&quot;alternatives-considered&quot;&gt;Alternatives considered&lt;/h2&gt;
&lt;p&gt;&lt;strong&gt;Keep review cycles PRIMARY and have each consumer resolve each fact from its own
home.&lt;/strong&gt; This is the alternative that actually competes, and the first draft omitted
it. It is not hypothetical: &lt;code dir=&quot;auto&quot;&gt;review_artifact_consistency.py&lt;/code&gt; already does exactly
this — &lt;code dir=&quot;auto&quot;&gt;_resolve_lane_state_read_dir&lt;/code&gt; + &lt;code dir=&quot;auto&quot;&gt;_resolve_review_cycle_read_dir&lt;/code&gt;, two kinds,
two homes, one caller. Because the gate already re-resolves both partitions from
mission identity, it does &lt;strong&gt;not&lt;/strong&gt; require co-location. Rejected on the operator’s
adjudication that per-WP review bookkeeping belongs on the shared coord surface as a
matter of where the trace should live, not because the gate cannot cope — the
motivation is reviewer-facing, and this ADR should not be read as claiming the gate
was broken without it.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Amend the glossary to say PRIMARY.&lt;/strong&gt; Rejected. Review cycles borrowed
&lt;code dir=&quot;auto&quot;&gt;WORK_PACKAGE_TASK&lt;/code&gt; because they live under &lt;code dir=&quot;auto&quot;&gt;tasks/&lt;/code&gt;, a &lt;em&gt;path&lt;/em&gt; coincidence rather
than a partition argument.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Leave both and let each consumer choose.&lt;/strong&gt; Rejected outright — the multi-authority
shape the charter’s single-canonical-authority principle exists to prevent.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Route by topology at each call site rather than by kind.&lt;/strong&gt; Rejected. It reproduces
the routing decision at every consumer, which is how the divergence arose.&lt;/p&gt;
&lt;h2 id=&quot;scope-note&quot;&gt;Scope note&lt;/h2&gt;
&lt;p&gt;&lt;code dir=&quot;auto&quot;&gt;notes&lt;/code&gt; and &lt;code dir=&quot;auto&quot;&gt;move-task&lt;/code&gt; also appear in the glossary’s COORD list and also have no
&lt;code dir=&quot;auto&quot;&gt;MissionArtifactKind&lt;/code&gt;. This ADR does not address them. Singling out review cycles is
a scoping decision driven by the referencing mission, not a derivation from P-1.&lt;/p&gt;
&lt;h2 id=&quot;references&quot;&gt;References&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;Glossary: &lt;a href=&quot;/context/orchestration/#coord-partition&quot;&gt;&lt;code dir=&quot;auto&quot;&gt;docs/context/orchestration.md#coord-partition&lt;/code&gt;&lt;/a&gt;, &lt;a href=&quot;/context/orchestration/#primary-partition&quot;&gt;&lt;code dir=&quot;auto&quot;&gt;#primary-partition&lt;/code&gt;&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href=&quot;/adr/3.x/2026-06-24-1-kind-and-topology-aware-artifact-placement/&quot;&gt;ADR 2026-06-24-1 — kind and topology aware artifact placement&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href=&quot;/adr/3.x/2026-07-19-1-wp-runtime-state-event-log-eviction-via-innerstatechanged/&quot;&gt;ADR 2026-07-19-1 — WP runtime state event-log eviction via InnerStateChanged&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;Referencing mission: &lt;code dir=&quot;auto&quot;&gt;kitty-specs/review-cycle-verdict-seam-rebuild-01KZ2W7W/&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;Code: &lt;code dir=&quot;auto&quot;&gt;src/mission_runtime/artifacts.py&lt;/code&gt;, &lt;code dir=&quot;auto&quot;&gt;src/specify_cli/coordination/commit_router.py&lt;/code&gt;, &lt;code dir=&quot;auto&quot;&gt;src/specify_cli/post_merge/review_artifact_consistency.py&lt;/code&gt;, &lt;code dir=&quot;auto&quot;&gt;src/specify_cli/merge/bookkeeping_projection.py&lt;/code&gt;&lt;/li&gt;
&lt;/ul&gt;</content>
</entry>
