chore(orchestrator): closure-sweep — C3 n/a-done.md fix + 18 verified umbrella closures (28 false-opens cleared)

.claude/orchestrator-prompts/ai-doc-audit-version-reconcile/done.md

@@ -0,0 +1,4 @@
+# ai-doc-audit-version-reconcile — DONE
+- Final PR: #423
+- Closed: 2026-06-05
+- Summary: closed via #423 — ai-doc-audit-version-reconcile: re-audit branches by audit's own criteria, take best

.claude/orchestrator-prompts/aif-handoff-as-runtime-bridge/done.md

@@ -0,0 +1,4 @@
+# aif-handoff-as-runtime-bridge — DONE
+- Final PR: #283
+- Closed: 2026-05-29
+- Summary: closed via #283 — docs(wave-plan): close aif-handoff-as-runtime-bridge umbrella row to DONE

.claude/orchestrator-prompts/aif-skills-delivery-rphase/done.md

@@ -0,0 +1,4 @@
+# aif-skills-delivery-rphase — DONE
+- Final PR: #396
+- Closed: 2026-06-03
+- Summary: closed via #396 — feat(aif-worker-discipline): condensed orchestrator-planning + reviewer layers in portable worker agent

.claude/orchestrator-prompts/channel-earliness-audit/done.md

@@ -0,0 +1,4 @@
+# channel-earliness-audit — DONE
+- Final PR: #181
+- Closed: 2026-05-22
+- Summary: closed via #181 — docs(channel-audit): channel-earliness retroactive sweep (§6-deferred)

.claude/orchestrator-prompts/defer-reflex-detection/done.md

@@ -0,0 +1,4 @@
+# defer-reflex-detection — DONE
+- Final PR: #232
+- Closed: 2026-05-25
+- Summary: closed via #232 — research(defer-reflex-detection): Stage 2 benchmark — REJECT UMBRELLA verdict

.claude/orchestrator-prompts/dispatcher-skill/done.md

@@ -0,0 +1,4 @@
+# dispatcher-skill — DONE
+- Final PR: #403
+- Closed: 2026-06-03
+- Summary: closed via #403 — feat(dispatcher): /dispatcher skill — wire aif-control execution loop (Stage 1)

.claude/orchestrator-prompts/dispatcher-ux/done.md

@@ -0,0 +1,4 @@
+# dispatcher-ux — DONE
+- Final PR: #407
+- Closed: 2026-06-03
+- Summary: closed via #407 — fix(dispatcher): P2 closure-marker + P3 base-norm + P4 self-apply + P6 watch-link (Stage 2)

.claude/orchestrator-prompts/goal-drift-audit/done.md

@@ -0,0 +1,4 @@
+# goal-drift-audit — DONE
+- Final PR: #433
+- Closed: 2026-06-05
+- Summary: closed via #433 — docs(goal-drift-audit): verdict patch + principle-09 criteria 3&4

.claude/orchestrator-prompts/guard-liveness-gate/done.md

@@ -0,0 +1,4 @@
+# guard-liveness-gate — DONE
+- Final PR: #460
+- Closed: 2026-06-11
+- Summary: closed via #460 — feat(guard-liveness): change-scoped ESLint guard-liveness gate (v1)

.claude/orchestrator-prompts/guard-liveness-stage2a-prelude/done.md

@@ -0,0 +1,4 @@
+# guard-liveness-stage2a-prelude — DONE
+- Final PR: #463
+- Closed: 2026-06-11
+- Summary: closed via #463 — feat(guard-liveness): Stage 2a prelude — fixture + pressure-scenario schema/types/principle

.claude/orchestrator-prompts/guard-liveness-v0-audit/done.md

@@ -0,0 +1,4 @@
+# guard-liveness-v0-audit — DONE
+- Final PR: #458
+- Closed: 2026-06-11
+- Summary: closed via #458 — research(guard-liveness-v0-audit): per-rule criticality table for 26 manifest rules

.claude/orchestrator-prompts/guard-liveness-v1-advisories/done.md

@@ -0,0 +1,4 @@
+# guard-liveness-v1-advisories — DONE
+- Final PR: #489
+- Closed: 2026-06-13
+- Summary: closed via #489 — chore(orchestrator): close guard-liveness umbrella (done.md)

.claude/orchestrator-prompts/guard-liveness-v1.5-cmd-script/done.md

@@ -0,0 +1,4 @@
+# guard-liveness-v1.5-cmd-script — DONE
+- Final PR: #475
+- Closed: 2026-06-13
+- Summary: closed via #475 — feat(guard-liveness): cmd/script rule liveness gate (v1.5)

.claude/orchestrator-prompts/guard-liveness-v3-manual-sp/done.md

@@ -0,0 +1,4 @@
+# guard-liveness-v3-manual-sp — DONE
+- Final PR: #476
+- Closed: 2026-06-13
+- Summary: closed via #476 — feat(guard-liveness): v3 manual-rule SP-prober (reduced, consolidated) — pressure-scenario schema + prober + 5 migrations

.claude/orchestrator-prompts/memory-coverage-audit/done.md

@@ -0,0 +1,4 @@
+# memory-coverage-audit — DONE
+- Final PR: #126
+- Closed: 2026-05-21
+- Summary: closed via #126 — docs(research): memory coverage audit (memory → docs → tests) 2026-05-22

.claude/orchestrator-prompts/meta-orchestrator-bundle-autonomous/done.md

@@ -0,0 +1,4 @@
+# meta-orchestrator-bundle-autonomous — DONE
+- Final PR: #247
+- Closed: 2026-05-26
+- Summary: closed via #247 — feat(meta-orchestrator): bundle-autonomous Stage 3 — B1 helper + B2 persistence + B3a self-eval

.claude/orchestrator-prompts/mutation-discipline-stage-c-meta-launch/done.md

@@ -0,0 +1,4 @@
+# mutation-discipline-stage-c-meta-launch — DONE
+- Final PR: #378
+- Closed: 2026-06-02
+- Summary: closed via #378 — mutation-discipline-stage-c-meta-launch

.claude/orchestrator-prompts/narrow-b-benchmark/done.md

@@ -0,0 +1,4 @@
+# narrow-b-benchmark — DONE
+- Final PR: #210
+- Closed: 2026-05-24
+- Summary: closed via #210 — research(narrow-b-benchmark): production-corpus FP measurement for Option B

.claude/skills/pipeline/helpers/priority-score.sh

@@ -222,7 +222,12 @@ for dir in "${PROMPTS_DIR}"/*/; do
   if [[ -z "${done_pr}" && -f "${dir}done.md" ]]; then
     done_pr="$(grep -m1 '^- Final PR: #' "${dir}done.md" 2>/dev/null \
       | grep -oE '[0-9]+' | head -n1 || true)"
-    [[ -n "${done_pr}" ]] && done_basis="done-md"
+    # done.md existence ALONE proves closure (C3 ADAPT Cline #77). A stale/superseded closure
+    # carries "Final PR: n/a" (no numeric PR) — still DONE; tag done_pr=n/a so the filter drops
+    # it. Pre-fix bug: n/a closures (e.g. the 2026-06-05 sweep) stayed false-open because this
+    # layer only tagged DONE on a numeric PR — fixed 2026-06-16 closure-sweep.
+    [[ -z "${done_pr}" ]] && done_pr="n/a"
+    done_basis="done-md"
   fi
 
   # Emit candidate line with optional DONE tag (basis indicates which layer matched).

packages/core/hooks/done-md-completion-filter.test.ts

@@ -309,6 +309,32 @@ describe('done-md-completion-filter — Layer C2 + C3 paired-negative contract',
     expect(line).not.toMatch(/basis=/);
   });
 
+  it('Case 3b — C3 n/a: done.md with "Final PR: n/a" → status=DONE done_pr=n/a basis=done-md', () => {
+    // Regression for the 2026-06-16 closure-sweep fix. A stale/superseded closure carries
+    // "- Final PR: n/a" (no numeric PR). done.md existence ALONE proves closure, so C3 MUST
+    // still tag DONE. Pre-fix bug: such n/a closures (e.g. the 2026-06-05 sweep) stayed
+    // false-open because the layer only tagged DONE when it parsed a numeric PR.
+    const sandbox = makeSandbox();
+    setupRepo(sandbox, ['na-done-umbrella', 'active-umbrella']);
+    const dir = join(sandbox, '.claude', 'orchestrator-prompts', 'na-done-umbrella');
+    mkdirSync(dir, { recursive: true });
+    writeFileSync(
+      join(dir, 'done.md'),
+      `# na-done-umbrella — DONE\n- Final PR: n/a (superseded)\n- Closed: 2026-06-05\n- Summary: scope absorbed elsewhere\n`,
+      'utf8',
+    );
+    const fakeGh = makeFakeGhEmpty(sandbox);
+    const fakeDup = makeFakeDupDetect(sandbox, {});
+
+    const r = runHelper(sandbox, fakeGh, { MO_DUP_DETECT_BIN: fakeDup });
+    expect(r.status).toBe(0);
+    // na-done-umbrella has done.md (n/a PR) → still DONE via Layer C3
+    expect(r.stdout).toMatch(/na-done-umbrella .* status=DONE done_pr=n\/a basis=done-md/);
+    // active-umbrella has no done.md → still candidate (paired-negative)
+    const activeLine = r.stdout.split('\n').find((l) => l.startsWith('active-umbrella '));
+    expect(activeLine).not.toMatch(/status=DONE/);
+  });
+
   it('Case 5 — COMBINED: C1 fail + C2 hit → DONE; C1 fail + C2 fail + C3 hit → DONE', () => {
     // Verifies first-match-wins ordering: when C1 misses but C2 or C3 matches, DONE is emitted.
     // Also verifies that C3 still fires when both C1 and C2 miss.