diff --git a/plugins/review-suite/scripts/review_suite_core/orchestrator_runner.py b/plugins/review-suite/scripts/review_suite_core/orchestrator_runner.py
index fbdc533..c750836 100644
--- a/plugins/review-suite/scripts/review_suite_core/orchestrator_runner.py
+++ b/plugins/review-suite/scripts/review_suite_core/orchestrator_runner.py
@@ -1450,6 +1450,8 @@ def _run_followup_review_once(
             f"{note} The reviewed head is no longer an ancestor of HEAD; review the current branch diff "
             f"against {base} and focus on whether the fixes address that source round's findings."
         )
+    if review_instructions := _review_instructions(state, {}):
+        note = f"{note}\n\n{review_instructions}"
     prompt = build_followup_prompt(
         since_head=since_head,
         head=head,
diff --git a/plugins/review-suite/scripts/review_suite_core/review_branch_status.py b/plugins/review-suite/scripts/review_suite_core/review_branch_status.py
index e4023a8..110dfbf 100644
--- a/plugins/review-suite/scripts/review_suite_core/review_branch_status.py
+++ b/plugins/review-suite/scripts/review_suite_core/review_branch_status.py
@@ -36,7 +36,6 @@
     public_task_name,
     read_jsonl,
     round_needs_caller_grade,
-    terminal_review_command,
     unique_round_state_dirs,
 )
 
@@ -605,8 +604,6 @@ def _round_terminal_command(round_record: dict[str, object]) -> str | None:
         if status and status != "completed":
             return None
         command = str(run.get("terminal_command") or "").strip().lower()
-        if not command:
-            command = terminal_review_command(str(run.get("reviewer_output") or ""))
         if command not in DECISION_COMMANDS:
             return None
         commands.append(command)
@@ -726,13 +723,24 @@ def _orchestrator_action(
                 }
             else:
                 action = {
-                    "cmd": _review_command(public_id, extra=("--decision", "clean")),
-                    "alt": _review_command(public_id, extra=("--decision", "findings")),
+                    "choices": {
+                        decision: _review_command(
+                            public_id, extra=("--decision", decision)
+                        )
+                        for decision in ("clean", "findings")
+                    },
+                    "note": "Classify the reviewer output, then record clean or findings.",
                 }
     elif stage == "fix-pending":
+        note = "Commit/amend valid fixes, then rerun this command."
+        if str(state.get("review_brief") or "").strip():
+            note += (
+                " If a finding conflicts with the frozen contract, rerun this review "
+                "id with --contract-conflict <dimension> instead."
+            )
         action = {
             "cmd": _review_command(public_id),
-            "note": "Commit/amend valid fixes, then rerun this command.",
+            "note": note,
         }
     elif stage in {"review-green", "local-green-handoff"}:
         summary = review_ladder_summary(state, current_head=current_head)
diff --git a/plugins/review-suite/tests/test_review_orchestrator_runner.py b/plugins/review-suite/tests/test_review_orchestrator_runner.py
index c0cd6fd..b6f40a6 100644
--- a/plugins/review-suite/tests/test_review_orchestrator_runner.py
+++ b/plugins/review-suite/tests/test_review_orchestrator_runner.py
@@ -1430,6 +1430,7 @@ def test_runner_runs_real_followup_once_from_followup_pending(
     fixed["identity"]["requested_base"] = "main"
     fixed["identity"]["base_upstream"] = "origin/main"
     fixed["identity"]["base_ref_stale"] = True
+    fixed["review_brief"] = "# Goal\n\nStay scoped."
 
     result = orchestrator_runner.run_one_expensive_step(
         fixed, state_dir=tmp_path / "state"
@@ -1468,6 +1469,7 @@ def test_runner_runs_real_followup_once_from_followup_pending(
     assert "Source review round phase_review-round-1" in str(
         followup_calls[0]["prompt"]
     )
+    _assert_review_brief_instructions(followup_calls[0]["prompt"])
 
 
 def test_runner_runs_rewritten_followup_against_branch_scope(
diff --git a/plugins/review-suite/tests/test_review_state.py b/plugins/review-suite/tests/test_review_state.py
index fa7339c..f73418f 100644
--- a/plugins/review-suite/tests/test_review_state.py
+++ b/plugins/review-suite/tests/test_review_state.py
@@ -622,16 +622,21 @@ def test_review_state_status_surfaces_orchestrator_progress(
     assert emitted[0]["progress"] == "review 1/2 broad-discovery"
     assert "recommendation" not in emitted[0]
     assert "reason" not in emitted[0]
-    assert "--id rvw_progress --decision clean" in str(emitted[0]["Action"]["cmd"])
-    assert "--id rvw_progress --decision findings" in str(emitted[0]["Action"]["alt"])
+    action = emitted[0]["Action"]
+    assert set(action) == {"choices", "note", "restart"}
+    assert action["note"] == (
+        "Classify the reviewer output, then record clean or findings."
+    )
+    assert "--id rvw_progress --decision clean" in str(action["choices"]["clean"])
+    assert "--id rvw_progress --decision findings" in str(action["choices"]["findings"])
     assert "--id rvw_progress --restart-mode deep --reason REASON" in str(
         emitted[0]["Action"]["restart"]["cmd"]
     )
     assert emitted[0]["Action"]["restart"]["mode"] == "deep"
-    assert "--state-dir" not in str(emitted[0]["Action"]["cmd"])
-    assert "--state-dir" not in str(emitted[0]["Action"]["alt"])
+    assert "--state-dir" not in str(action["choices"]["clean"])
+    assert "--state-dir" not in str(action["choices"]["findings"])
     assert "--state-dir" not in str(emitted[0]["Action"]["restart"]["cmd"])
-    assert str(state_dir.resolve(strict=False)) not in str(emitted[0]["Action"]["cmd"])
+    assert str(state_dir.resolve(strict=False)) not in str(action["choices"]["clean"])
     assert "review_t1.py" not in str(emitted[0]["Action"])
 
 
@@ -773,7 +778,8 @@ def test_review_state_status_uses_bare_id_for_structured_verdict(
                     {
                         "slot": "alpha",
                         "review_status": "completed",
-                        "reviewer_output": "No findings.\n\nReview result: clean",
+                        "reviewer_output": "Review result: findings",
+                        "terminal_command": "clean",
                         "grade_blocked": False,
                     }
                 ],
@@ -818,6 +824,16 @@ def test_review_state_status_uses_bare_id_for_structured_verdict(
     assert "--id rvw_progress --decision findings" in str(
         action["override"]["findings"]
     )
+    briefed_fix = review_state._orchestrator_action(
+        {"stage": "fix-pending", "review_brief": "Frozen contract"},
+        "rvw_progress",
+        state_dir=state_dir,
+    )
+    briefless_fix = review_state._orchestrator_action(
+        {"stage": "fix-pending"}, "rvw_progress", state_dir=state_dir
+    )
+    assert "--contract-conflict <dimension>" in str(briefed_fix["note"])
+    assert briefless_fix["note"] == "Commit/amend valid fixes, then rerun this command."
 
 
 def test_review_state_status_surfaces_grade_before_structured_verdict(