diff --git a/README.md b/README.md
index a065e76..9921b10 100644
--- a/README.md
+++ b/README.md
@@ -385,6 +385,18 @@ Règles modélisées :
   coût Dijkstra depuis la cible, diagonales 5-10-5), en s'arrêtant adjacent à
   l'ennemi le plus proche — les combattants contournent les murs au lieu
   d'osciller contre eux.
+  Destination explicite : `Action(kind="move", destination=(r, c))` et
+  `Action(kind="5foot_step", destination=(r, c))` déplacent le combattant vers
+  la case `(r, c)` désignée (monter au couvert, quitter un défilé, se
+  désengager) au lieu de auto-diriger vers `target_id`. Pour `move`, la case
+  doit être praticable, libre et atteignable dans le budget de vitesse (sinon
+  rejet `unreachable_destination`) ; le trajet réutilise la mécanique pas à pas
+  avec provocation d'AoO. Pour `5foot_step`, la case doit être exactement
+  adjacente (sinon `unreachable_destination`) et le pas ne provoque jamais
+  d'AoO. Sans `destination`, le comportement auto-dirigé est inchangé ; avec
+  `target_id` et `destination` absents, rejet `missing_target` (U1).
+  `default_policy` ne choisit pas de destination (disponible via une politique
+  personnalisée).
 - Tailles et occupation multi-cases : une créature occupe un footprint carré à
   son ancre (`pos`) selon sa taille — Small/Medium 1×1, Large 2×2, Huge 3×3,
   Gargantuan 4×4, Colossal 6×6 (taille inconnue → 1×1). Distance, ligne
diff --git a/src/pf1e_simulator/combat.py b/src/pf1e_simulator/combat.py
index 6fdf029..fa3098a 100644
--- a/src/pf1e_simulator/combat.py
+++ b/src/pf1e_simulator/combat.py
@@ -540,7 +540,10 @@ class Action:
     ``fight_defensively`` also takes -4 on its attacks. ``stand_up`` is a move
     action ending the prone condition (it provokes an AoO from each enemy
     threatening the combatant); ``drop_prone`` is a free action going prone
-    until the combatant stands up.
+    until the combatant stands up. ``move`` and ``5foot_step`` accept an
+    explicit ``destination`` cell ``(row, col)``: with one they walk to that
+    cell instead of auto-pathing toward ``target_id`` (the 5-foot step's
+    destination must be exactly one cell away).
     """
 
     kind: Literal[
@@ -563,6 +566,7 @@ class Action:
     trick_condition: TrickCondition | None = None  # dirty trick condition choice
     weapon_id: str | None = None  # weapon id for retrieve_weapon / repair_weapon
     ready_trigger: ReadyTrigger | None = None  # ready action trigger event
+    destination: tuple[int, int] | None = None  # explicit (row, col) cell for move / 5foot_step
 
 
 # Normalized action-result reason vocabulary. U1 emits ok, unsupported_kind,
@@ -570,13 +574,14 @@ class Action:
 # no_line_of_effect, no_slot and insufficient_resource are declared here for
 # the deeper execution-feedback package (U6b). U3 adds delayed /
 # already_delayed for the delay action. U4 adds ready for the ready action.
-# U5b adds not_prone / already_prone for the prone actions.
+# U5b adds not_prone / already_prone for the prone actions. U5c adds
+# unreachable_destination for move / 5foot_step with an explicit destination.
 ACTION_RESULT_REASONS: frozenset[str] = frozenset({
     "ok", "unsupported_kind", "missing_target", "no_weapon", "out_of_range",
     "no_line_of_effect", "no_slot", "insufficient_resource",
     "downgraded:staggered", "downgraded:nauseated",
     "delayed", "already_delayed", "ready",
-    "not_prone", "already_prone",
+    "not_prone", "already_prone", "unreachable_destination",
 })
 _REASON_OK = "ok"
 _REASON_UNSUPPORTED_KIND = "unsupported_kind"
@@ -589,12 +594,14 @@ _REASON_ALREADY_DELAYED = "already_delayed"
 _REASON_READY = "ready"
 _REASON_NOT_PRONE = "not_prone"
 _REASON_ALREADY_PRONE = "already_prone"
+_REASON_UNREACHABLE_DESTINATION = "unreachable_destination"
 
 # Action kinds that are declared but have no engine dispatch (U1 surfaces
 # them as explicit rejections instead of silent no-ops).
 _UNSUPPORTED_ACTION_KINDS: frozenset[str] = frozenset({"swift", "free", "full_round"})
 # Action kinds whose helper is a silent no-op (or worse) without a resolved
 # target; _execute rejects them up front with missing_target instead.
+# move / 5foot_step are exempt when an explicit destination is provided (U5c).
 _TARGET_REQUIRED_ACTION_KINDS: frozenset[str] = frozenset({
     "attack", "full_attack", "charge", "withdraw", "move", "5foot_step",
     "cast_spell", "maneuver",
@@ -2430,20 +2437,37 @@ class CombatEngine:
                 kind=kind, ok=False, reason=_REASON_UNSUPPORTED_KIND
             )
         target = next((s for s in self._states if s.combatant.id == action.target_id), None)
-        if target is None and kind in _TARGET_REQUIRED_ACTION_KINDS:
-            self._log(
-                f"round {self._current_round} {state.combatant.id}: "
-                f"{kind} rejected ({_REASON_MISSING_TARGET})"
-            )
-            return ActionResult(kind=kind, ok=False, reason=_REASON_MISSING_TARGET)
+        rejection = self._target_rejection(state, kind, target, action)
+        if rejection is not None:
+            return rejection
         if kind == "delay":
             return self._execute_delay(state, target, action)
         if kind == "ready":
             return self._execute_ready(state, action)
         if kind in ("total_defense", "fight_defensively", "stand_up", "drop_prone"):
             return self._execute_self_action(state, kind)
-        self._dispatch_action(state, target, action)
-        return ActionResult(kind=kind, ok=True, reason=_REASON_OK)
+        override = self._dispatch_action(state, target, action)
+        return override if override is not None else ActionResult(
+            kind=kind, ok=True, reason=_REASON_OK
+        )
+
+    def _target_rejection(
+        self,
+        state: CombatantState,
+        kind: str,
+        target: CombatantState | None,
+        action: Action,
+    ) -> ActionResult | None:
+        """U1 missing-target gate; move/5foot_step accept a destination instead (U5c)."""
+        if target is not None or kind not in _TARGET_REQUIRED_ACTION_KINDS:
+            return None
+        if kind in ("move", "5foot_step") and action.destination is not None:
+            return None
+        self._log(
+            f"round {self._current_round} {state.combatant.id}: "
+            f"{kind} rejected ({_REASON_MISSING_TARGET})"
+        )
+        return ActionResult(kind=kind, ok=False, reason=_REASON_MISSING_TARGET)
 
     def _execute_delay(
         self, state: CombatantState, target: CombatantState | None, action: Action
@@ -2597,8 +2621,13 @@ class CombatEngine:
 
     def _dispatch_action(
         self, state: CombatantState, target: CombatantState | None, action: Action
-    ) -> None:
-        """Route a validated action to its execution helper."""
+    ) -> ActionResult | None:
+        """Route a validated action to its execution helper.
+
+        Returns an explicit rejection when a destination-bearing movement
+        cannot be honoured (U5c); None means the helper ran and the caller
+        reports ok.
+        """
         kind = action.kind
         if kind == "cast_spell":
             self._execute_cast_spell(state, target, action.spell_name)
@@ -2622,7 +2651,8 @@ class CombatEngine:
         elif kind == "immediate":
             self._execute_immediate(state, target, action)
         else:
-            self._execute_movement(state, target, kind)
+            return self._execute_movement(state, target, action)
+        return None
 
     def _execute_class_ability(
         self, state: CombatantState, target: CombatantState | None, action: Action
@@ -2643,21 +2673,32 @@ class CombatEngine:
             self._execute_inspire_courage(state, target)
 
     def _execute_movement(
-        self, state: CombatantState, target: CombatantState | None, kind: str
-    ) -> None:
-        """Execute a movement-style action (5foot_step, move, wait)."""
+        self, state: CombatantState, target: CombatantState | None, action: Action
+    ) -> ActionResult | None:
+        """Execute a movement-style action (5foot_step, move, wait).
+
+        With an explicit destination, move/5foot_step walk to that cell and
+        report unreachable_destination when it cannot be honoured (U5c);
+        without one they auto-path toward the target as before.
+        """
+        kind = action.kind
         if kind == "5foot_step":
+            if action.destination is not None:
+                return self._step5_to(state, action.destination)
             if target is None:
-                return
+                return None
             self._step5(state, target)
         elif kind == "move":
+            if action.destination is not None:
+                return self._move_to(state, action.destination)
             if target is None:
-                return
+                return None
             self._move(state, target)
         elif kind == "wait":
             self._log(f"round {self._current_round} {state.combatant.id}: wait")
         elif kind == "activate_power_attack":
             self._execute_activate_power_attack(state)
+        return None
 
     def _execute_immediate(
         self, state: CombatantState, target: CombatantState | None, action: Action
@@ -4145,6 +4186,33 @@ class CombatEngine:
             f"5ft step ({start[0]},{start[1]})->({path[0][0]},{path[0][1]})"
         )
 
+    def _step5_to(self, state: CombatantState, destination: Pos) -> ActionResult:
+        """5-foot step to an explicit adjacent cell — never provokes (U5c)."""
+        kind = "5foot_step"
+        if state.moved_this_turn:
+            return ActionResult(kind=kind, ok=True, reason=_REASON_OK)
+        if self.has_condition_flag(state, "cant_move"):
+            self._log(f"round {self._current_round} {state.combatant.id}: wait (cannot move)")
+            return ActionResult(kind=kind, ok=True, reason=_REASON_OK)
+        adjacent = self._grid.distance(state.pos, destination) == 1
+        free = self._cells_free(self._footprint_at(destination, state), state)
+        if not adjacent or not free:
+            self._log(
+                f"round {self._current_round} {state.combatant.id}: "
+                f"{kind} rejected ({_REASON_UNREACHABLE_DESTINATION})"
+            )
+            return ActionResult(
+                kind=kind, ok=False, reason=_REASON_UNREACHABLE_DESTINATION
+            )
+        start = state.pos
+        state.pos = destination
+        state.moved_this_turn = True
+        self._log(
+            f"round {self._current_round} {state.combatant.id}: "
+            f"5ft step ({start[0]},{start[1]})->({destination[0]},{destination[1]})"
+        )
+        return ActionResult(kind=kind, ok=True, reason=_REASON_OK)
+
     def _move(self, state: CombatantState, target: CombatantState) -> None:
         if state.moved_this_turn:
             return
@@ -4153,6 +4221,58 @@ class CombatEngine:
             return
         path = self._move_path(state, target)
         start = state.pos
+        actual_path = self._walk_steps(state, path)
+        if actual_path:
+            state.moved_this_turn = True
+            coords = "->".join(f"({p[0]},{p[1]})" for p in (start, *actual_path))
+            self._log(f"round {self._current_round} {state.combatant.id}: move {coords}")
+
+    def _move_to(self, state: CombatantState, destination: Pos) -> ActionResult:
+        """Move action to an explicit cell within speed (U5c).
+
+        Validated against the Dijkstra cost field from the combatant's position
+        under its speed budget; walked step-by-step with the same AoO
+        machinery as the auto-pathed move.
+        """
+        kind = "move"
+        if state.moved_this_turn:
+            return ActionResult(kind=kind, ok=True, reason=_REASON_OK)
+        if self.has_condition_flag(state, "cant_move"):
+            self._log(f"round {self._current_round} {state.combatant.id}: wait (cannot move)")
+            return ActionResult(kind=kind, ok=True, reason=_REASON_OK)
+        if not self._destination_reachable(state, destination):
+            self._log(
+                f"round {self._current_round} {state.combatant.id}: "
+                f"{kind} rejected ({_REASON_UNREACHABLE_DESTINATION})"
+            )
+            return ActionResult(
+                kind=kind, ok=False, reason=_REASON_UNREACHABLE_DESTINATION
+            )
+        path = self._descend_path(state, destination)
+        self._walk_steps(state, path)
+        state.moved_this_turn = True
+        self._log(
+            f"round {self._current_round} {state.combatant.id}: "
+            f"move -> ({destination[0]},{destination[1]})"
+        )
+        return ActionResult(kind=kind, ok=True, reason=_REASON_OK)
+
+    def _destination_reachable(self, state: CombatantState, destination: Pos) -> bool:
+        """True when destination is passable, unoccupied and within this speed."""
+        speed_cells = int(
+            state.combatant.speed_land_ft // _SQUARE_FT * self._get_speed_mult(state)
+        )
+        if speed_cells <= 0:
+            return destination == state.pos
+        if not self._cells_free(self._footprint_at(destination, state), state):
+            return False
+        field = self._grid.reachable(state.pos, speed_cells, self._blocked(state))
+        return destination in field
+
+    def _walk_steps(
+        self, state: CombatantState, path: list[Pos]
+    ) -> list[Pos]:
+        """Walk a precomputed path cell by cell, resolving AoOs and reactions."""
         actual_path: list[Pos] = []
         for step in path:
             if self._check_move_aoo(state):
@@ -4164,10 +4284,7 @@ class CombatEngine:
             self._check_reactions(ReactionEvent(kind="movement_step", actor=state))
             if not state.active:
                 break
-        if actual_path:
-            state.moved_this_turn = True
-            coords = "->".join(f"({p[0]},{p[1]})" for p in (start, *actual_path))
-            self._log(f"round {self._current_round} {state.combatant.id}: move {coords}")
+        return actual_path
 
     def _move_path(self, state: CombatantState, target: CombatantState) -> list[Pos]:
         """Full movement path toward the target, up to the combatant's speed.
@@ -4175,12 +4292,16 @@ class CombatEngine:
         Greedy walk of the anchor over the Dijkstra cost field from the target,
         requiring the whole footprint to fit at each step.
         """
+        return self._descend_path(state, target.pos)
+
+    def _descend_path(self, state: CombatantState, anchor: Pos) -> list[Pos]:
+        """Greedy walk of the anchor over the cost field from ``anchor``."""
         speed_mult = self._get_speed_mult(state)
         speed_cells = int(state.combatant.speed_land_ft // _SQUARE_FT * speed_mult)
         if speed_cells <= 0:
             return []
         blocked = self._blocked(state)
-        to_target = self._grid.reachable(target.pos, None, blocked)
+        to_target = self._grid.reachable(anchor, None, blocked)
         if self._footprint_cost(state.occupied_cells(), to_target) is None:
             return []
         path: list[Pos] = []
diff --git a/tests/test_move_destination.py b/tests/test_move_destination.py
new file mode 100644
index 0000000..821c001
--- /dev/null
+++ b/tests/test_move_destination.py
@@ -0,0 +1,289 @@
+"""Explicit movement destinations (Package U5c).
+
+``move`` and ``5foot_step`` accept ``Action.destination`` — a ``(row, col)``
+cell. With a destination the combatant walks to THAT cell (validated against
+the Dijkstra cost field under its speed budget, with the usual occupied-cell
+blocking) instead of auto-pathing toward ``target_id``; the step-by-step walk
+reuses the same AoO machinery as the auto-pathed move. A ``5foot_step``
+destination must be exactly one cell away and never provokes. Rejections
+(out of bounds, impassable, occupied, beyond speed, non-adjacent) report
+``unreachable_destination``. Without a destination both kinds behave exactly
+as before, and the U1 ``missing_target`` rule still applies when neither a
+target nor a destination is given.
+"""
+from __future__ import annotations
+
+from collections.abc import Callable
+
+from pf1e_simulator.combat import (
+    ACTION_RESULT_REASONS,
+    Action,
+    ActionResult,
+    CombatantState,
+    CombatEngine,
+)
+from pf1e_simulator.dice import parse_dice
+from pf1e_simulator.grid import Grid
+from pf1e_simulator.map import MapSpec, TerrainType
+from pf1e_simulator.models import (
+    AbilityScores,
+    ACProfile,
+    AttackSpec,
+    Combatant,
+    DamageComponent,
+    Saves,
+)
+from pf1e_simulator.rng import ScriptedRng
+from tests.test_observation_enriched import make_grid, make_state
+
+PolicyFn = Callable[[CombatEngine, CombatantState], tuple[Action, ...]]
+
+_WAIT: tuple[Action, ...] = (Action(kind="wait"),)
+
+
+def _combatant(id_: str, *, initiative_mod: int = 0) -> Combatant:
+    return Combatant(
+        id=id_,
+        name=id_,
+        level=1,
+        abilities=AbilityScores(
+            str_score=10, dex_score=10, con_score=10,
+            int_score=10, wis_score=10, cha_score=10,
+        ),
+        ac=ACProfile(total=10, touch=10, flat_footed=10),
+        bab=1,
+        initiative_mod=initiative_mod,
+        speed_land_ft=30,
+        hp_max=100,
+        saves=Saves(fort=0, ref=0, will=0),
+        attacks=[
+            AttackSpec(
+                id=f"{id_}-sword",
+                name="short sword",
+                kind="melee",
+                attack_bonus=5,
+                damage=[
+                    DamageComponent(formula=parse_dice("1d6"), types=["slashing"])
+                ],
+            )
+        ],
+    )
+
+
+def _wall_grid() -> Grid:
+    """Open 8x8 floor with a wall column segment at (0, 2)."""
+    rows = [".." + "#" + "." * 5] + ["." * 8] * 7
+    legend = {
+        ".": TerrainType(type="floor", move_cost=1),
+        "#": TerrainType(type="wall", move_cost=None, blocks_los=True),
+    }
+    spec = MapSpec(name="walls", terrain=tuple(rows), legend=legend)
+    return Grid.from_spec(spec)
+
+
+def _hero_vs_foe(
+    policy: PolicyFn, rng: list[int], round_cap: int, *, foe_pos: tuple[int, int],
+    grid: Grid | None = None,
+) -> CombatEngine:
+    """hero (acts first) at (0, 0); foe at ``foe_pos``; scripted rolls."""
+    hero = make_state(_combatant("hero", initiative_mod=2), "players", (0, 0))
+    foe = make_state(_combatant("foe"), "monsters", foe_pos)
+    return CombatEngine(
+        ScriptedRng(rng), grid or make_grid(), [hero, foe],
+        policy=policy, round_cap=round_cap,
+    )
+
+
+# ── (a) move to a specific free cell off the auto-path ────────────────────────
+
+
+def test_move_to_destination_off_the_enemy_path() -> None:
+    # Foe at (0, 4): the auto-path would walk hero along row 0 toward (0, 3).
+    # The destination (3, 0) is NOT on that path — hero must go there instead.
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="move", destination=(3, 0)),)
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5], round_cap=1, foe_pos=(0, 4))
+    t = eng.run().transcript
+    hero = eng._states[0]
+    assert hero.pos == (3, 0)
+    assert "round 1 hero: move -> (3,0)" in t
+    assert not any("move (0,0)->" in line for line in t)
+    assert hero.last_action_results == [
+        ActionResult(kind="move", ok=True, reason="ok")
+    ]
+
+
+def test_move_destination_walk_uses_aoo_machinery() -> None:
+    # Foe at (1, 1) threatens (0, 0): the first step away provokes an AoO
+    # (d20 1 + 5 = 6 vs AC 10 -> MISS) resolved before leaving the square.
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="move", destination=(3, 0)),)
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5, 1], round_cap=1, foe_pos=(1, 1))
+    t = eng.run().transcript
+    assert eng._states[0].pos == (3, 0)
+    assert "round 1 foe: AoO vs hero d20=1+5=6 AC 10 -> MISS" in t
+    lines = "\n".join(t)
+    assert lines.index("AoO vs hero") < lines.index("round 1 hero: move -> (3,0)")
+    assert eng._aoo_used == {"foe": 1}
+
+
+# ── (b) out-of-bounds / impassable / occupied destinations ────────────────────
+
+
+def test_out_of_bounds_destination_is_unreachable() -> None:
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="move", destination=(99, 99)),)
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5], round_cap=1, foe_pos=(0, 4))
+    t = eng.run().transcript
+    assert "round 1 hero: move rejected (unreachable_destination)" in t
+    assert eng._states[0].pos == (0, 0)
+    assert eng._states[0].last_action_results == [
+        ActionResult(kind="move", ok=False, reason="unreachable_destination")
+    ]
+    assert "unreachable_destination" in ACTION_RESULT_REASONS
+
+
+def test_impassable_destination_is_unreachable() -> None:
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="move", destination=(0, 2)),)
+        return _WAIT
+
+    eng = _hero_vs_foe(
+        policy, [10, 5], round_cap=1, foe_pos=(0, 4), grid=_wall_grid()
+    )
+    t = eng.run().transcript
+    assert "round 1 hero: move rejected (unreachable_destination)" in t
+    assert eng._states[0].pos == (0, 0)
+
+
+def test_occupied_destination_is_unreachable() -> None:
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="move", destination=(0, 4)),)
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5], round_cap=1, foe_pos=(0, 4))
+    t = eng.run().transcript
+    assert "round 1 hero: move rejected (unreachable_destination)" in t
+    assert eng._states[0].pos == (0, 0)
+
+
+# ── (c) destination beyond the speed budget ───────────────────────────────────
+
+
+def test_destination_beyond_speed_budget_is_unreachable() -> None:
+    # speed 30 ft = 6 cells; (0, 7) is 7 cells away — out of reach this turn.
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="move", destination=(0, 7)),)
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5], round_cap=1, foe_pos=(7, 7))
+    t = eng.run().transcript
+    assert "round 1 hero: move rejected (unreachable_destination)" in t
+    assert eng._states[0].pos == (0, 0)
+    assert eng._states[0].last_action_results == [
+        ActionResult(kind="move", ok=False, reason="unreachable_destination")
+    ]
+
+
+# ── (d) 5-foot step to an adjacent cell; non-adjacent rejected ────────────────
+
+
+def test_step5_to_adjacent_destination_does_not_provoke() -> None:
+    # Foe at (0, 1) threatens the starting square: the 5-ft step must NOT
+    # provoke (no extra scripted rolls beyond initiative).
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="5foot_step", destination=(1, 0)),)
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5], round_cap=1, foe_pos=(0, 1))
+    t = eng.run().transcript
+    hero = eng._states[0]
+    assert hero.pos == (1, 0)
+    assert "round 1 hero: 5ft step (0,0)->(1,0)" in t
+    assert not any("AoO" in line for line in t)
+    assert hero.last_action_results == [
+        ActionResult(kind="5foot_step", ok=True, reason="ok")
+    ]
+
+
+def test_step5_to_non_adjacent_destination_is_rejected() -> None:
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="5foot_step", destination=(0, 2)),)
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5], round_cap=1, foe_pos=(0, 4))
+    t = eng.run().transcript
+    assert "round 1 hero: 5foot_step rejected (unreachable_destination)" in t
+    assert eng._states[0].pos == (0, 0)
+    assert eng._states[0].last_action_results == [
+        ActionResult(kind="5foot_step", ok=False, reason="unreachable_destination")
+    ]
+
+
+def test_step5_to_occupied_destination_is_rejected() -> None:
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="5foot_step", destination=(0, 1)),)
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5], round_cap=1, foe_pos=(0, 1))
+    t = eng.run().transcript
+    assert "round 1 hero: 5foot_step rejected (unreachable_destination)" in t
+    assert eng._states[0].pos == (0, 0)
+
+
+# ── (e) U1 preserved: neither target nor destination → missing_target ─────────
+
+
+def test_move_without_target_or_destination_is_missing_target() -> None:
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="move"), Action(kind="5foot_step"))
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5], round_cap=1, foe_pos=(0, 4))
+    t = eng.run().transcript
+    assert "round 1 hero: move rejected (missing_target)" in t
+    assert "round 1 hero: 5foot_step rejected (missing_target)" in t
+    assert eng._states[0].last_action_results == [
+        ActionResult(kind="move", ok=False, reason="missing_target"),
+        ActionResult(kind="5foot_step", ok=False, reason="missing_target"),
+    ]
+
+
+# ── (f) target-only move still auto-paths byte-identically ────────────────────
+
+
+def test_move_with_target_auto_paths_as_before() -> None:
+    def policy(_engine: CombatEngine, state: CombatantState) -> tuple[Action, ...]:
+        if state.combatant.id == "hero":
+            return (Action(kind="move", target_id="foe"),)
+        return _WAIT
+
+    eng = _hero_vs_foe(policy, [10, 5], round_cap=1, foe_pos=(0, 5))
+    result = eng.run()
+    assert result.transcript == (
+        "initiative: hero d20=10+2=12",
+        "initiative: foe d20=5+0=5",
+        "round 1 hero: move (0,0)->(0,1)->(0,2)->(0,3)->(0,4)",
+        "round 1 foe: wait",
+        "battle over: draw after 1 rounds",
+    )
+    assert eng._states[0].pos == (0, 4)
+    assert eng._states[0].last_action_results == [
+        ActionResult(kind="move", ok=True, reason="ok")
+    ]