perf(goal-channel): defer repository audit on refresh writeback

docs/architecture/rfcs/loopx-overall-roadmap-v0.md

@@ -237,6 +237,18 @@ expansion follows from this roadmap revision.
 
 Do not invent unmeasured performance targets. Before each experiment/pilot, its owner freezes thresholds, baseline, budget, stop conditions and evidence scope. A post-result threshold change belongs to a new experiment. G4 correctness requires no duplicate protected effects or stale/unauthorized commits; performance and cost thresholds require separate measured qualification.
 
+Operational status used by post-writeback sinks must stay distinct from repository
+publication audits. A Goal Channel gate decision still reads current Goal and
+quota state, but a repository-wide public-boundary scan belongs to explicit
+`loopx check`/premerge validation, not every `refresh-state` notification
+attempt. The local status APIs already use this separation. For the next
+long-history qualification, measure the full `refresh-state` and `quota
+spend-slot` paths separately: lock wait, history/receipt readback, state
+projection, optional sink work, and result delivery. Keep the required
+authority and privacy checks; optimize repeated reads only with equivalent
+positive, negative, retry, and stale-generation outcomes. A fast sink no-op
+is not evidence that long-history status and quota admission meet G4 SLOs.
+
 
 ## 4. Every RFC: Ownership and Next Step
 

loopx/extensions/lark/goal_channel_lifecycle.py

@@ -197,6 +197,10 @@ def sync_human_gate_after_refresh(
             scan_roots=[registry_project_root(source_registry_path)],
             limit=20,
             goal_id=goal_id,
+            # Gate delivery needs current Goal/quota state, not a repository-wide
+            # publication audit on every committed refresh. Keep that audit in
+            # explicit check/premerge flows, as the local status APIs do.
+            include_public_boundary_scan=False,
         )
         quota_packet = build_quota_should_run(
             status,

tests/extensions/test_lark_goal_channel_lifecycle.py

@@ -207,10 +207,16 @@ def test_refresh_lifecycle_no_gate_checks_extension_without_delivery(
 ) -> None:
     registry_path = _registry(tmp_path)
     _binding(registry_path, enabled=True)
+    status_reads: list[dict[str, Any]] = []
+
+    def operational_status(**kwargs: Any) -> dict[str, Any]:
+        status_reads.append(kwargs)
+        return {"status": "fixture"}
+
     monkeypatch.setattr(
         goal_channel_lifecycle,
         "collect_status",
-        lambda **kwargs: {"status": "fixture"},
+        operational_status,
     )
     monkeypatch.setattr(
         goal_channel_lifecycle,
@@ -236,6 +242,9 @@ def test_refresh_lifecycle_no_gate_checks_extension_without_delivery(
     assert result["enabled"] is True
     assert result["status"] == "not_selected"
     assert result["extension_activation"] == {"status": "active"}
+    assert len(status_reads) == 1
+    assert status_reads[0]["goal_id"] == GOAL_ID
+    assert status_reads[0]["include_public_boundary_scan"] is False
 
 
 def test_refresh_lifecycle_extension_failure_prevents_private_binding_read(