diff --git a/adventure/actions/base.py b/adventure/actions/base.py
index 86d174d..cde29d7 100644
--- a/adventure/actions/base.py
+++ b/adventure/actions/base.py
@@ -7,6 +7,7 @@ from adventure.context import (
     get_agent_for_actor,
     world_context,
 )
+from adventure.errors import ActionError
 from adventure.utils.conversation import loop_conversation
 from adventure.utils.search import (
     find_actor_in_room,
@@ -79,11 +80,11 @@ def action_move(direction: str) -> str:
             None,
         )
         if not portal:
-            return f"You cannot move {direction} from here."
+            raise ActionError(f"You cannot move {direction} from here.")
 
         destination_room = find_room(action_world, portal.destination)
         if not destination_room:
-            return f"The {portal.destination} room does not exist."
+            raise ActionError(f"The {portal.destination} room does not exist.")
 
         broadcast(
             f"{action_actor.name} moves through {direction} to {destination_room.name}"
@@ -106,7 +107,7 @@ def action_take(item: str) -> str:
     with action_context() as (action_room, action_actor):
         action_item = find_item_in_room(action_room, item)
         if not action_item:
-            return f"The {item} item is not in the room."
+            raise ActionError(f"The {item} item is not in the room.")
 
         broadcast(f"{action_actor.name} takes the {item} item")
         action_room.items.remove(action_item)
@@ -127,13 +128,15 @@ def action_ask(character: str, question: str) -> str:
         # sanity checks
         question_actor, question_agent = get_actor_agent_for_name(character)
         if question_actor == action_actor:
-            return "You cannot ask yourself a question. Stop talking to yourself. Try another action."
+            raise ActionError(
+                "You cannot ask yourself a question. Stop talking to yourself. Try another action."
+            )
 
         if not question_actor:
-            return f"The {character} character is not in the room."
+            raise ActionError(f"The {character} character is not in the room.")
 
         if not question_agent:
-            return f"The {character} character does not exist."
+            raise ActionError(f"The {character} character does not exist.")
 
         broadcast(f"{action_actor.name} asks {character}: {question}")
         first_prompt = (
@@ -183,13 +186,15 @@ def action_tell(character: str, message: str) -> str:
         # sanity checks
         question_actor, question_agent = get_actor_agent_for_name(character)
         if question_actor == action_actor:
-            return "You cannot tell yourself a message. Stop talking to yourself. Try another action."
+            raise ActionError(
+                "You cannot tell yourself a message. Stop talking to yourself. Try another action."
+            )
 
         if not question_actor:
-            return f"The {character} character is not in the room."
+            raise ActionError(f"The {character} character is not in the room.")
 
         if not question_agent:
-            return f"The {character} character does not exist."
+            raise ActionError(f"The {character} character does not exist.")
 
         broadcast(f"{action_actor.name} tells {character}: {message}")
         first_prompt = (
@@ -236,11 +241,16 @@ def action_give(character: str, item: str) -> str:
     with action_context() as (action_room, action_actor):
         destination_actor = find_actor_in_room(action_room, character)
         if not destination_actor:
-            return f"The {character} character is not in the room."
+            raise ActionError(f"The {character} character is not in the room.")
+
+        if destination_actor == action_actor:
+            raise ActionError(
+                "You cannot give an item to yourself. Try another action."
+            )
 
         action_item = find_item_in_actor(action_actor, item)
         if not action_item:
-            return f"You do not have the {item} item in your inventory."
+            raise ActionError(f"You do not have the {item} item in your inventory.")
 
         broadcast(f"{action_actor.name} gives {character} the {item} item.")
         action_actor.items.remove(action_item)
@@ -260,7 +270,7 @@ def action_drop(item: str) -> str:
     with action_context() as (action_room, action_actor):
         action_item = find_item_in_actor(action_actor, item)
         if not action_item:
-            return f"You do not have the {item} item in your inventory."
+            raise ActionError(f"You do not have the {item} item in your inventory.")
 
         broadcast(f"{action_actor.name} drops the {item} item")
         action_actor.items.remove(action_item)
diff --git a/adventure/actions/optional.py b/adventure/actions/optional.py
index af6e754..97acad1 100644
--- a/adventure/actions/optional.py
+++ b/adventure/actions/optional.py
@@ -13,6 +13,7 @@ from adventure.context import (
     set_dungeon_master,
     world_context,
 )
+from adventure.errors import ActionError
 from adventure.generate import generate_item, generate_room, link_rooms
 from adventure.utils.effect import apply_effects
 from adventure.utils.search import find_actor_in_room
@@ -47,7 +48,10 @@ def action_explore(direction: str) -> str:
 
         if direction in action_room.portals:
             dest_room = action_room.portals[direction]
-            return f"You cannot explore {direction} from here, that direction already leads to {dest_room}. Please use the move action to go there."
+            raise ActionError(
+                f"You cannot explore {direction} from here, that direction already leads to {dest_room}. "
+                "Please use the move action to go there."
+            )
 
         try:
             systems = get_game_systems()
@@ -118,7 +122,7 @@ def action_use(item: str, target: str) -> str:
             None,
         )
         if not action_item:
-            return f"The {item} item is not available to use."
+            raise ActionError(f"The {item} item is not available to use.")
 
         if target == "self":
             target_actor = action_actor
@@ -148,13 +152,15 @@ def action_use(item: str, target: str) -> str:
         )
         if not chosen_effect:
             # TODO: should retry the question if the effect is not found
-            return f"The {chosen_name} effect is not available to apply."
+            raise ValueError(f"The {chosen_name} effect is not available to apply.")
 
         try:
             apply_effects(target_actor, [chosen_effect])
         except Exception:
             logger.exception("error applying effect: %s", chosen_effect)
-            return f"There was a problem applying the {chosen_name} effect."
+            raise ValueError(
+                f"There was a problem applying the {chosen_name} effect while using the {item} item."
+            )
 
         broadcast(
             f"{action_actor.name} uses the {chosen_name} effect of {item} on {target}"
diff --git a/adventure/errors.py b/adventure/errors.py
new file mode 100644
index 0000000..a5d5b4c
--- /dev/null
+++ b/adventure/errors.py
@@ -0,0 +1,2 @@
+class ActionError(Exception):
+    pass
diff --git a/adventure/generate.py b/adventure/generate.py
index b59cad3..c1bf52a 100644
--- a/adventure/generate.py
+++ b/adventure/generate.py
@@ -1,5 +1,5 @@
 from logging import getLogger
-from random import choice, randint
+from random import choice
 from typing import List, Tuple
 
 from packit.agent import Agent
@@ -19,6 +19,7 @@ from adventure.models.effect import (
 from adventure.models.entity import Actor, Item, Portal, Room, World, WorldEntity
 from adventure.models.event import GenerateEvent
 from adventure.utils import try_parse_float, try_parse_int
+from adventure.utils.effect import resolve_int_range
 from adventure.utils.search import (
     list_actors,
     list_actors_in_room,
@@ -108,9 +109,7 @@ def generate_room(
     actions = {}
     room = Room(name=name, description=desc, items=[], actors=[], actions=actions)
 
-    item_count = randint(
-        world_config.size.room_items.min, world_config.size.room_items.max
-    )
+    item_count = resolve_int_range(world_config.size.room_items) or 0
     broadcast_generated(f"Generating {item_count} items for room: {name}")
 
     for _ in range(item_count):
@@ -127,9 +126,7 @@ def generate_room(
         except Exception:
             logger.exception("error generating item")
 
-    actor_count = randint(
-        world_config.size.room_actors.min, world_config.size.room_actors.max
-    )
+    actor_count = resolve_int_range(world_config.size.room_actors) or 0
     broadcast_generated(message=f"Generating {actor_count} actors for room: {name}")
 
     for _ in range(actor_count):
@@ -265,9 +262,7 @@ def generate_item(
     item = Item(name=name, description=desc, actions=actions)
     generate_system_attributes(agent, world, item, systems)
 
-    effect_count = randint(
-        world_config.size.item_effects.min, world_config.size.item_effects.max
-    )
+    effect_count = resolve_int_range(world_config.size.item_effects) or 0
     broadcast_generated(message=f"Generating {effect_count} effects for item: {name}")
 
     for _ in range(effect_count):
@@ -326,9 +321,7 @@ def generate_actor(
     generate_system_attributes(agent, world, actor, systems)
 
     # generate the actor's inventory
-    item_count = randint(
-        world_config.size.actor_items.min, world_config.size.actor_items.max
-    )
+    item_count = resolve_int_range(world_config.size.actor_items) or 0
     broadcast_generated(f"Generating {item_count} items for actor {name}")
 
     for k in range(item_count):
@@ -470,9 +463,7 @@ def link_rooms(
     rooms = rooms or world.rooms
 
     for room in rooms:
-        num_portals = randint(
-            world_config.size.portals.min, world_config.size.portals.max
-        )
+        num_portals = resolve_int_range(world_config.size.portals) or 0
 
         if len(room.portals) >= num_portals:
             logger.info(f"room {room.name} already has enough portals")
@@ -517,9 +508,7 @@ def generate_world(
     systems: List[GameSystem],
     room_count: int | None = None,
 ) -> World:
-    room_count = room_count or randint(
-        world_config.size.rooms.min, world_config.size.rooms.max
-    )
+    room_count = room_count or resolve_int_range(world_config.size.rooms) or 0
 
     broadcast_generated(message=f"Generating a {theme} with {room_count} rooms")
     world = World(name=name, rooms=[], theme=theme, order=[])
diff --git a/adventure/models/config.py b/adventure/models/config.py
index 5f05914..7e70364 100644
--- a/adventure/models/config.py
+++ b/adventure/models/config.py
@@ -22,11 +22,11 @@ class BotConfig:
 
 @dataclass
 class RenderConfig:
-    cfg: IntRange
+    cfg: int | IntRange
     checkpoints: List[str]
     path: str
     sizes: Dict[str, Size]
-    steps: IntRange
+    steps: int | IntRange
 
 
 @dataclass
@@ -47,12 +47,12 @@ class WorldActorConfig:
 
 @dataclass
 class WorldSizeConfig:
-    actor_items: IntRange
-    item_effects: IntRange
-    portals: IntRange
-    room_actors: IntRange
-    room_items: IntRange
-    rooms: IntRange
+    actor_items: int | IntRange
+    item_effects: int | IntRange
+    portals: int | IntRange
+    room_actors: int | IntRange
+    room_items: int | IntRange
+    rooms: int | IntRange
 
 
 @dataclass
@@ -87,11 +87,11 @@ DEFAULT_CONFIG = Config(
     server=ServerConfig(websocket=WebsocketServerConfig(host="localhost", port=8001)),
     world=WorldConfig(
         actor=WorldActorConfig(
-            conversation_limit=3,
+            conversation_limit=2,
         ),
         size=WorldSizeConfig(
             actor_items=IntRange(min=0, max=2),
-            item_effects=IntRange(min=1, max=2),
+            item_effects=IntRange(min=1, max=1),
             portals=IntRange(min=1, max=3),
             rooms=IntRange(min=3, max=6),
             room_actors=IntRange(min=1, max=3),
diff --git a/adventure/render/comfy.py b/adventure/render/comfy.py
index ab9ce29..42c6310 100644
--- a/adventure/render/comfy.py
+++ b/adventure/render/comfy.py
@@ -28,6 +28,7 @@ from adventure.models.event import (
     ResultEvent,
     StatusEvent,
 )
+from adventure.utils.random import resolve_int_range
 
 from .prompt import prompt_from_entity, prompt_from_event
 
@@ -44,17 +45,11 @@ render_thread: Thread | None = None
 
 
 def generate_cfg():
-    if render_config.cfg.min == render_config.cfg.max:
-        return render_config.cfg.min
-
-    return randint(render_config.cfg.min, render_config.cfg.max)
+    return resolve_int_range(render_config.cfg)
 
 
 def generate_steps():
-    if render_config.steps.min == render_config.steps.max:
-        return render_config.steps.min
-
-    return randint(render_config.steps.min, render_config.steps.max)
+    return resolve_int_range(render_config.steps)
 
 
 def generate_batches(
diff --git a/adventure/simulate.py b/adventure/simulate.py
index d5aea04..e4c59dd 100644
--- a/adventure/simulate.py
+++ b/adventure/simulate.py
@@ -1,5 +1,6 @@
 from functools import partial
 from itertools import count
+from json import loads
 from logging import getLogger
 from math import inf
 from typing import Callable, Sequence
@@ -91,6 +92,22 @@ def prompt_actor_action(
         if not room or not actor:
             raise ValueError("Room and actor must be set before parsing results")
 
+        # trim suffixes that are used elsewhere
+        value = value.removesuffix("END").strip()
+
+        # fix unbalanced curly braces
+        if value.startswith("{") and not value.endswith("}"):
+            open_count = value.count("{")
+            close_count = value.count("}")
+
+            if open_count > close_count:
+                fixed_value = value + ("}" * (open_count - close_count))
+                try:
+                    loads(fixed_value)
+                    value = fixed_value
+                except Exception:
+                    pass
+
         if could_be_json(value):
             event = ActionEvent.from_json(value, room, actor)
         else:
diff --git a/adventure/utils/conversation.py b/adventure/utils/conversation.py
index 5bdacee..a2d96e1 100644
--- a/adventure/utils/conversation.py
+++ b/adventure/utils/conversation.py
@@ -11,6 +11,7 @@ from packit.utils import could_be_json
 from adventure.context import broadcast
 from adventure.models.config import DEFAULT_CONFIG
 from adventure.models.entity import Actor, Room
+from adventure.models.event import ReplyEvent
 
 from .string import normalize_name
 
@@ -125,10 +126,12 @@ def loop_conversation(
     if len(actors) != len(agents):
         raise ValueError("The number of actors and agents must match.")
 
+    # set up the keyword or length-limit compound condition
     _, condition_end, parse_end = make_keyword_condition(end_message)
     stop_length = partial(condition_threshold, max=max_length)
     stop_condition = condition_or(condition_end, stop_length)
 
+    # prepare a result parser looking for the echo function
     def result_parser(value: str, **kwargs) -> str:
         value = parse_end(value, **kwargs)
 
@@ -140,6 +143,7 @@ def loop_conversation(
 
         return value.strip()
 
+    # prepare the loop state
     i = 0
     last_actor = first_actor
     response = first_message
@@ -163,8 +167,9 @@ def loop_conversation(
         )
         response = result_parser(response)
 
-        logger.info(f"{actor.name} response: {response}")
-        broadcast(f"{actor.name} responds to {last_actor.name}: {response}")
+        logger.info(f"{actor.name} responds: {response}")
+        reply_event = ReplyEvent.from_text(response, room, actor)
+        broadcast(reply_event)
 
         # increment the step counter
         i += 1
diff --git a/adventure/utils/effect.py b/adventure/utils/effect.py
index f38435f..5a25690 100644
--- a/adventure/utils/effect.py
+++ b/adventure/utils/effect.py
@@ -1,8 +1,6 @@
-import random
 from logging import getLogger
 from typing import List
 
-from adventure.models.base import FloatRange, IntRange
 from adventure.models.effect import (
     BooleanEffectPattern,
     BooleanEffectResult,
@@ -23,6 +21,8 @@ from adventure.utils.attribute import (
     prepend_value,
 )
 
+from .random import resolve_float_range, resolve_int_range, resolve_string_list
+
 logger = getLogger(__name__)
 
 
@@ -124,50 +124,6 @@ def effective_attributes(
     return attributes
 
 
-def resolve_float_range(range: float | FloatRange | None) -> float | None:
-    """
-    Resolve a float range to a single value.
-    """
-
-    if range is None:
-        return None
-
-    if isinstance(
-        range, (float, int)
-    ):  # int is not really necessary here, but mypy complains without it
-        return range
-
-    return random.uniform(range.min, range.max)
-
-
-def resolve_int_range(range: int | IntRange | None) -> int | None:
-    """
-    Resolve an integer range to a single value.
-    """
-
-    if range is None:
-        return None
-
-    if isinstance(range, int):
-        return range
-
-    return random.randint(range.min, range.max)
-
-
-def resolve_string_list(result: str | List[str] | None) -> str | None:
-    """
-    Resolve a string result to a single value.
-    """
-
-    if result is None:
-        return None
-
-    if isinstance(result, str):
-        return result
-
-    return random.choice(result)
-
-
 def resolve_boolean_effect(effect: BooleanEffectPattern) -> BooleanEffectResult:
     """
     Apply a boolean effect pattern to a set of attributes.
diff --git a/adventure/utils/random.py b/adventure/utils/random.py
new file mode 100644
index 0000000..7ee924e
--- /dev/null
+++ b/adventure/utils/random.py
@@ -0,0 +1,48 @@
+import random
+from typing import List
+
+from adventure.models.base import FloatRange, IntRange
+
+
+def resolve_float_range(range: float | FloatRange | None) -> float | None:
+    """
+    Resolve a float range to a single value.
+    """
+
+    if range is None:
+        return None
+
+    if isinstance(
+        range, (float, int)
+    ):  # int is not really necessary here, but mypy complains without it
+        return range
+
+    return random.uniform(range.min, range.max)
+
+
+def resolve_int_range(range: int | IntRange | None) -> int | None:
+    """
+    Resolve an integer range to a single value.
+    """
+
+    if range is None:
+        return None
+
+    if isinstance(range, int):
+        return range
+
+    return random.randint(range.min, range.max)
+
+
+def resolve_string_list(result: str | List[str] | None) -> str | None:
+    """
+    Resolve a string result to a single value.
+    """
+
+    if result is None:
+        return None
+
+    if isinstance(result, str):
+        return result
+
+    return random.choice(result)