diff --git a/prod/clean_arch/violet/cadence.py b/prod/clean_arch/violet/cadence.py
new file mode 100644
index 0000000..21f9c85
--- /dev/null
+++ b/prod/clean_arch/violet/cadence.py
@@ -0,0 +1,170 @@
+"""VIOLET V3b: Cadence Control Plane — per-action quantization.
+
+Spec: prod/docs/VIOLET_SPEC__CADENCE_CONTROL_PLANE.md (BINDING).
+
+VIOLET is a reactor substrate (V0 clock/DeadlineScheduler); BLUE's scan-quantized
+behaviour is a guest hosted on it. Scan cadence is a QUANTIZATION SETTING, not the
+architecture. Every scan-governed action is *evaluated* every reactor tick (the
+would-be action shadow-logged) and *actuated* only when its per-action quantum Q is
+crossed. Each action's Q is:
+
+  - UNIVERSAL      — every action (entry, sizing, each exit reason, each input plane)
+                     has one; nothing is hardcoded at scan. SL merely DEFAULTS tighter.
+  - INDEPENDENT    — changing one action's Q never touches another's.
+  - CONTROL-PLANE  — readable/writable at runtime (HZ-backed via refresh_from); code
+                     defaults are the floor when the control plane is silent.
+  - LOOSENABLE     — Q steps down per-action on its own schedule, each step gated on
+                     the shadow deltas (evaluate-vs-actuate) this layer makes visible.
+
+Exit-priority (CATASTROPHIC/ADVSL > mechanical TP > discretionary v7) is INVARIANT to
+Q and is enforced by the decision engine, not here — a faster Q never lets a
+discretionary exit mask a mechanical one (the LINK -$1,248 bug).
+"""
+
+from __future__ import annotations
+
+import enum
+from dataclasses import dataclass
+from typing import Callable, Dict, Optional
+
+# ── quantum constants (nanoseconds) ───────────────────────────────────────────
+INSTA_Q_NS = 0                  # actuate every reactor tick (no quantization)
+OBF_Q_NS = 1_000_000_000        # ~1s — OBF effective cadence (fastest service)
+SCAN_Q_NS = 5_000_000_000       # 5s — NG7 scan cadence (the certification anchor)
+
+
+class Action(str, enum.Enum):
+    """Every scan-governed VIOLET action that carries an independent cadence knob."""
+
+    CATASTROPHIC_SL = "catastrophic_sl"
+    ADVSL = "advsl"
+    TP = "tp"
+    V7_EXIT = "v7_exit"
+    ENTRY = "entry"
+    SIZING = "sizing"
+    CONSUME_SCAN = "consume_scan"
+    CONSUME_OBF = "consume_obf"
+    CONSUME_EXF = "consume_exf"
+    CONSUME_ESOF = "consume_esof"
+    CONSUME_MARAS = "consume_maras"
+    CONSUME_ACB = "consume_acb"
+
+
+# Default Q per action (spec §3). SL-class defaults to insta (the safety deviation);
+# everything else defaults to its source cadence — all loosenable independently.
+_DEFAULT_Q_NS: Dict[Action, int] = {
+    Action.CATASTROPHIC_SL: INSTA_Q_NS,
+    Action.ADVSL: INSTA_Q_NS,
+    Action.TP: SCAN_Q_NS,
+    Action.V7_EXIT: SCAN_Q_NS,
+    Action.ENTRY: SCAN_Q_NS,
+    Action.SIZING: SCAN_Q_NS,
+    Action.CONSUME_SCAN: SCAN_Q_NS,
+    Action.CONSUME_OBF: OBF_Q_NS,
+    Action.CONSUME_EXF: SCAN_Q_NS,
+    Action.CONSUME_ESOF: SCAN_Q_NS,
+    Action.CONSUME_MARAS: SCAN_Q_NS,
+    Action.CONSUME_ACB: SCAN_Q_NS,
+}
+
+
+@dataclass
+class CadenceKnob:
+    """One action's tunable cadence. Mutable — the control plane is its only mutator.
+
+    ``q_ns`` is the actuation quantum: 0 ⇒ actuate every reactor tick (insta); else
+    actuate once ``q_ns`` has elapsed since the last actuation. ``evaluate_every_tick``
+    keeps the would-be action computed (and shadow-logged) at full reactor speed
+    regardless of Q.
+    """
+
+    action: Action
+    q_ns: int
+    evaluate_every_tick: bool = True
+    enabled: bool = True
+    source: str = "default"           # "default" | "control_plane"
+
+    def __post_init__(self) -> None:
+        if int(self.q_ns) < 0:
+            raise ValueError(f"q_ns must be >= 0 (got {self.q_ns} for {self.action})")
+        self.q_ns = int(self.q_ns)
+
+
+class CadenceControlPlane:
+    """Runtime registry of per-action cadence knobs (spec §4).
+
+    Seeded from code defaults; ``refresh_from`` layers a runtime provider (HZ map /
+    env) on top, per action, leaving untouched actions at their current value.
+    """
+
+    def __init__(self) -> None:
+        self._knobs: Dict[Action, CadenceKnob] = {
+            a: CadenceKnob(action=a, q_ns=q) for a, q in _DEFAULT_Q_NS.items()
+        }
+
+    def get(self, action: Action) -> CadenceKnob:
+        return self._knobs[action]
+
+    def set(
+        self, action: Action, *,
+        q_ns: Optional[int] = None,
+        evaluate_every_tick: Optional[bool] = None,
+        enabled: Optional[bool] = None,
+        source: str = "control_plane",
+    ) -> CadenceKnob:
+        """Independently override one action. Other actions are never touched."""
+        k = self._knobs[action]
+        new = CadenceKnob(
+            action=action,
+            q_ns=k.q_ns if q_ns is None else int(q_ns),
+            evaluate_every_tick=k.evaluate_every_tick if evaluate_every_tick is None else bool(evaluate_every_tick),
+            enabled=k.enabled if enabled is None else bool(enabled),
+            source=source,
+        )
+        self._knobs[action] = new
+        return new
+
+    def refresh_from(self, provider: Callable[[Action], Optional[dict]]) -> int:
+        """Pull runtime overrides from a provider (HZ map / env reader).
+
+        ``provider(action)`` returns a dict of overrides or None (no override → keep
+        current value, i.e. the code-default floor). Returns the count of actions
+        updated. The control plane stays authoritative for live tuning.
+        """
+        n = 0
+        for action in Action:
+            ov = provider(action)
+            if not ov:
+                continue
+            self.set(action, **{k: v for k, v in ov.items()
+                                if k in ("q_ns", "evaluate_every_tick", "enabled")})
+            n += 1
+        return n
+
+    def due(self, action: Action, now_ns: int, last_actuation_ns: Optional[int]) -> bool:
+        """Q-boundary test: should this action ACTUATE now?
+
+        Disabled ⇒ never. ``last_actuation_ns is None`` (never actuated) ⇒ yes.
+        q_ns == 0 (insta) ⇒ every tick. Else ⇒ once the quantum has elapsed.
+        """
+        k = self._knobs[action]
+        if not k.enabled:
+            return False
+        if last_actuation_ns is None:
+            return True
+        if k.q_ns == 0:
+            return True
+        return (int(now_ns) - int(last_actuation_ns)) >= k.q_ns
+
+    # Alias matching spec §4 caller idiom (evaluate always, then consult to actuate).
+    should_actuate = due
+
+    def snapshot(self) -> Dict[str, dict]:
+        """Surfaced view for live inspection (spec §4)."""
+        return {
+            a.value: {
+                "q_ns": k.q_ns, "evaluate_every_tick": k.evaluate_every_tick,
+                "enabled": k.enabled, "source": k.source,
+            }
+            for a, k in self._knobs.items()
+        }
diff --git a/prod/clean_arch/violet/test_violet_cadence.py b/prod/clean_arch/violet/test_violet_cadence.py
new file mode 100644
index 0000000..53093bc
--- /dev/null
+++ b/prod/clean_arch/violet/test_violet_cadence.py
@@ -0,0 +1,96 @@
+"""V3b: Cadence Control Plane — defaults, independence, control-plane, Q boundaries."""
+
+from __future__ import annotations
+
+import sys
+
+sys.path.insert(0, "/mnt/dolphinng5_predict")
+
+import pytest
+from hypothesis import given, settings, strategies as st
+
+from prod.clean_arch.violet.cadence import (
+    INSTA_Q_NS, OBF_Q_NS, SCAN_Q_NS, Action, CadenceControlPlane, CadenceKnob,
+)
+
+
+def test_defaults_match_spec_table():
+    cp = CadenceControlPlane()
+    assert cp.get(Action.CATASTROPHIC_SL).q_ns == INSTA_Q_NS
+    assert cp.get(Action.ADVSL).q_ns == INSTA_Q_NS
+    assert cp.get(Action.TP).q_ns == SCAN_Q_NS
+    assert cp.get(Action.CONSUME_OBF).q_ns == OBF_Q_NS
+    assert cp.get(Action.ENTRY).q_ns == SCAN_Q_NS
+    # SL-class defaults strictly tighter than TP (the safety deviation).
+    assert cp.get(Action.CATASTROPHIC_SL).q_ns < cp.get(Action.TP).q_ns
+    # every action is registered (universality) and evaluates every tick.
+    for a in Action:
+        assert cp.get(a).evaluate_every_tick is True
+
+
+def test_set_is_independent_per_action():
+    cp = CadenceControlPlane()
+    before = {a: cp.get(a).q_ns for a in Action}
+    cp.set(Action.TP, q_ns=1_000_000_000)
+    assert cp.get(Action.TP).q_ns == 1_000_000_000
+    assert cp.get(Action.TP).source == "control_plane"
+    # nothing else moved
+    for a in Action:
+        if a is Action.TP:
+            continue
+        assert cp.get(a).q_ns == before[a]
+        assert cp.get(a).source == "default"
+
+
+def test_control_plane_override_beats_default_absence_falls_back():
+    cp = CadenceControlPlane()
+    overrides = {Action.TP: {"q_ns": 250_000_000}, Action.ENTRY: {"enabled": False}}
+    n = cp.refresh_from(lambda a: overrides.get(a))
+    assert n == 2
+    assert cp.get(Action.TP).q_ns == 250_000_000          # overridden
+    assert cp.get(Action.ENTRY).enabled is False
+    assert cp.get(Action.SIZING).q_ns == SCAN_Q_NS         # untouched → default floor
+
+
+def test_negative_q_rejected_at_construction():
+    with pytest.raises(ValueError):
+        CadenceKnob(action=Action.TP, q_ns=-1)
+
+
+def test_due_insta_actuates_every_tick():
+    cp = CadenceControlPlane()
+    assert cp.due(Action.CATASTROPHIC_SL, now_ns=1000, last_actuation_ns=999) is True
+    # never-actuated always due
+    assert cp.due(Action.TP, now_ns=0, last_actuation_ns=None) is True
+
+
+def test_due_scan_respects_quantum():
+    cp = CadenceControlPlane()
+    last = 1_000_000_000
+    assert cp.due(Action.TP, now_ns=last + SCAN_Q_NS - 1, last_actuation_ns=last) is False
+    assert cp.due(Action.TP, now_ns=last + SCAN_Q_NS, last_actuation_ns=last) is True
+
+
+def test_disabled_action_never_due():
+    cp = CadenceControlPlane()
+    cp.set(Action.V7_EXIT, enabled=False)
+    assert cp.due(Action.V7_EXIT, now_ns=10**12, last_actuation_ns=None) is False
+
+
+@given(
+    q=st.integers(min_value=0, max_value=10_000_000_000),
+    elapsed=st.integers(min_value=0, max_value=20_000_000_000),
+)
+@settings(max_examples=100, deadline=None)
+def test_due_boundary_property(q, elapsed):
+    cp = CadenceControlPlane()
+    cp.set(Action.ENTRY, q_ns=q)
+    last = 5_000_000_000
+    due = cp.due(Action.ENTRY, now_ns=last + elapsed, last_actuation_ns=last)
+    assert due == (q == 0 or elapsed >= q)
+
+
+def test_snapshot_surfaces_all_actions():
+    snap = CadenceControlPlane().snapshot()
+    assert set(snap.keys()) == {a.value for a in Action}
+    assert all({"q_ns", "enabled", "source"} <= set(v) for v in snap.values())
diff --git a/prod/docs/VIOLET_SPEC__CADENCE_CONTROL_PLANE.md b/prod/docs/VIOLET_SPEC__CADENCE_CONTROL_PLANE.md
new file mode 100644
index 0000000..1aa3e44
--- /dev/null
+++ b/prod/docs/VIOLET_SPEC__CADENCE_CONTROL_PLANE.md
@@ -0,0 +1,97 @@
+# VIOLET Spec — Cadence Control Plane (per-action quantization)
+
+**Status:** BINDING for V3b (`prod/clean_arch/violet/cadence.py`). Authored 2026-06-13
+from operator doctrine this session. Companion to the master charter
+(memory `violet_subsecond_rebuild_plan`, cadence-quantizer section) and
+`violet_v3_alpha_doctrine` (points #6, #8). This spec is the painstaking per-ACTION
+quantization layer the charter only sketched as a principle.
+
+---
+
+## 1. The inversion (why this layer exists)
+
+BLUE's clock **is** its architecture: every decision/action is denominated in 5–6 s
+scans. VIOLET inverts this — the architecture is the **event-driven / clockless
+(fastest-clock) reactor** (V0: `clock.py` mono_ns timebase + `DeadlineScheduler`).
+BLUE's scan-quantized behaviour is a **guest** hosted on that reactor. The scan
+cadence is a **quantization setting, not the architecture.**
+
+At first, every action is quantized at **Q = scan cadence** → bit-faithful BLUE
+replication (warts and all). Over time each action's Q can be **loosened
+independently** toward the reactor's faster clock. This is the certification-preserving
+path: champion params are 5 s-bar-denominated (BIBLE §22.3), so loosening is a
+deliberate, measured, per-action promotion — never a global flip.
+
+## 2. Core rules
+
+1. **Evaluate at fastest cadence; actuate at Q.** Every action is *evaluated* every
+   reactor tick (the would-be action is computed and **shadow-logged** — this is the
+   evidence trail). The action is only *actuated* when its quantization boundary Q is
+   crossed. Shadow deltas (would-be-at-fast vs actuated-at-Q) are the data that
+   justifies each loosening step.
+2. **Adjustability is UNIVERSAL.** EVERY scan-governed activity carries its own
+   tunable Q — entry, sizing, each exit reason (TP, catastrophic SL, ADVSL, v7),
+   AND each input-plane consumption (scan, OBF, ExoF, EsoF, MARAS, ACB). No action is
+   hardcoded at scan. SL merely **defaults** to a tighter Q; mechanically it is the
+   same adjustable primitive as everything else.
+3. **Per-action knobs are INDEPENDENTLY configurable.** Changing TP's Q must not
+   touch entry's Q. Each action is its own registry entry.
+4. **Knobs are SURFACED IN A CONTROL PLANE.** The registry is readable and writable at
+   runtime for live inspection and tuning — not compile-time constants. HZ-backed at
+   runtime (mirroring how BLUE reads `DOLPHIN_FEATURES["live_tp_threshold"]`); code
+   defaults are the floor/fallback when the control plane is silent.
+5. **Loosening is per-reason and measured.** Q steps down per action on its own
+   schedule (e.g. 6s→3s→1s→500ms→100ms→50ms), each promotion gated on the shadow
+   deltas from rule (1). SL-class first; TP later (TP at scan has desirable properties).
+
+## 3. Per-action default Q table (initial state — all loosenable)
+
+| Action | Class | Default Q | Loosening intent |
+|---|---|---|---|
+| Catastrophic SL | exit/safety | **tight VIOLET / insta-exec** | sub-second from early (the versioned safety deviation) |
+| ADVSL | exit/safety | tight VIOLET (after min-bars gate) | sub-second early |
+| TP (mechanical) | exit | **scan** | loosen *cautiously* (desirable at scan) |
+| v7 discretionary exit | exit | scan | as-is first; per-reason later |
+| Entry | entry | scan | cautious (champion params 5s-denominated) |
+| Sizing (conviction) | sizing | scan | cautious (couples to entry) |
+| Scan plane consume | input | scan (~5–6s native) | n/a (source cadence) |
+| OBF plane consume | input | **~1s** (500ms purported / ~1s effective) | toward VIOLET clock — fastest service |
+| ExoF / EsoF / MARAS / ACB consume | input | scan | adjustable too, though sources are orders slower |
+
+Exit priority is INVARIANT regardless of Q: CATASTROPHIC/ADVSL > mechanical TP >
+discretionary (v7). A faster Q never lets a discretionary exit mask a mechanical one
+(the LINK −$1,248 bug).
+
+## 4. Design (`cadence.py`)
+
+- **`Action`** — enum/identifier per row of §3 (CATASTROPHIC_SL, ADVSL, TP, V7_EXIT,
+  ENTRY, SIZING, CONSUME_SCAN, CONSUME_OBF, CONSUME_EXF, CONSUME_ESOF, CONSUME_MARAS,
+  CONSUME_ACB).
+- **`CadenceKnob`** (V-TYPES `StrictModel`-style, but mutable via the control plane) —
+  per action: `q_ns` (actuation quantum; 0 ⇒ every reactor tick = "insta"),
+  `evaluate_every_tick: bool` (default True), `enabled: bool`, `source: "default"|"control_plane"`.
+- **`CadenceControlPlane`** — the registry: `get(action) -> CadenceKnob`,
+  `set(action, **overrides)` (independent per action), `snapshot() -> dict` (surfaced
+  for inspection), `refresh_from(provider)` where `provider` reads the HZ map / env at
+  runtime; code defaults seed it. `due(action, now_ns, last_actuation_ns) -> bool`
+  (Q-boundary test, integrates with `DeadlineScheduler`).
+- **Evaluate/actuate split helper** — `should_actuate(action, now_ns) -> bool`;
+  callers always evaluate, then consult this to actuate, and emit the shadow-delta
+  telemetry on the gap.
+
+## 5. Tests (V3b)
+
+- defaults match §3 table; SL defaults tighter than TP.
+- `set` on one action leaves all others unchanged (independence).
+- control-plane override beats default; absence falls back to default (floor).
+- `due`/`should_actuate` honors Q boundaries (insta Q=0 actuates every tick; scan Q
+  actuates once per scan interval), property-tested with hypothesis.
+- evaluate-always invariant: evaluation count ≥ actuation count for every action.
+- exit-priority invariant unaffected by Q (a fast discretionary Q never preempts a
+  mechanical exit in the ordering).
+
+## 6. Related
+
+`violet_subsecond_rebuild_plan` (charter) · `violet_v3_alpha_doctrine` #6/#8 ·
+`clock.py`/`DeadlineScheduler` (V0 substrate) · `decision_engine.py` (V3c consumer) ·
+`prod/docs/TODO_TP_SCAN_CADENCE_BUGFIX.md` (TP-at-scan rationale).