diff --git a/agents/agent_utils.py b/agents/agent_utils.py
index cde0afa..13c4e89 100644
--- a/agents/agent_utils.py
+++ b/agents/agent_utils.py
@@ -10,7 +10,7 @@
 from os import path
 sys.path.append(path.dirname(path.dirname(path.dirname( path.abspath(__file__) ))))
 #with the path fixed, we can import now
-from env.game_components import Action, ActionType, GameState, Observation, IP, Network
+from AIDojoCoordinator.game_components import Action, ActionType, GameState, Observation, IP, Network
 import ipaddress
 
 def generate_valid_actions_concepts(state: GameState)->list:
@@ -22,12 +22,12 @@ def generate_valid_actions_concepts(state: GameState)->list:
             # TODO ADD neighbouring networks
             # Only scan local networks from local hosts
             if network.is_private() and source_host.is_private():
-                valid_actions.add(Action(ActionType.ScanNetwork, params={"target_network": network, "source_host": source_host,}))
+                valid_actions.add(Action(ActionType.ScanNetwork, parameters={"target_network": network, "source_host": source_host,}))
         # Service Scans
         for host in state.known_hosts:
             # Do not try to scan a service from hosts outside local networks towards local networks
             if host.is_private() and source_host.is_private():
-                valid_actions.add(Action(ActionType.FindServices, params={"target_host": host, "source_host": source_host,}))
+                valid_actions.add(Action(ActionType.FindServices, parameters={"target_host": host, "source_host": source_host,}))
         # Service Exploits
         for host, service_list in state.known_services.items():
             # Only exploit local services from local hosts
@@ -37,52 +37,65 @@ def generate_valid_actions_concepts(state: GameState)->list:
                     for service in service_list:
                         # Do not consider local services, which are internal to the host
                         if not service.is_local:
-                            valid_actions.add(Action(ActionType.ExploitService, params={"target_host": host,"target_service": service,"source_host": source_host,}))
+                            valid_actions.add(Action(ActionType.ExploitService, parameters={"target_host": host,"target_service": service,"source_host": source_host,}))
     # Find Data Scans
     for host in state.controlled_hosts:
-        valid_actions.add(Action(ActionType.FindData, params={"target_host": host, "source_host": host}))
+        valid_actions.add(Action(ActionType.FindData, parameters={"target_host": host, "source_host": host}))
 
     # Data Exfiltration
     for source_host, data_list in state.known_data.items():
         for data in data_list:
             for target_host in state.controlled_hosts:
                 if target_host != source_host:
-                    valid_actions.add(Action(ActionType.ExfiltrateData, params={"target_host": target_host, "source_host": source_host, "data": data}))
+                    valid_actions.add(Action(ActionType.ExfiltrateData, parameters={"target_host": target_host, "source_host": source_host, "data": data}))
     return list(valid_actions)
 
-def generate_valid_actions(state: GameState)->list:
+def generate_valid_actions(state: GameState, include_blocks=False)->list:
     """Function that generates a list of all valid actions in a given state"""
     valid_actions = set()
+    def is_fw_blocked(state, src_ip, dst_ip)->bool:
+        blocked = False
+        try:
+            blocked = dst_ip in state.known_blocks[src_ip]
+        except KeyError:
+            pass #this src ip has no known blocks
+        return blocked 
+
     for src_host in state.controlled_hosts:
         #Network Scans
         for network in state.known_networks:
             # TODO ADD neighbouring networks
-            valid_actions.add(Action(ActionType.ScanNetwork, params={"target_network": network, "source_host": src_host,}))
+            valid_actions.add(Action(ActionType.ScanNetwork, parameters={"target_network": network, "source_host": src_host,}))
         # Service Scans
         for host in state.known_hosts:
-            valid_actions.add(Action(ActionType.FindServices, params={"target_host": host, "source_host": src_host,}))
+            if not is_fw_blocked(state, src_host,host):
+                valid_actions.add(Action(ActionType.FindServices, parameters={"target_host": host, "source_host": src_host,}))
         # Service Exploits
         for host, service_list in state.known_services.items():
-            for service in service_list:
-                valid_actions.add(Action(ActionType.ExploitService, params={"target_host": host,"target_service": service,"source_host": src_host,}))
+            if not is_fw_blocked(state, src_host,host):
+                for service in service_list:
+                    valid_actions.add(Action(ActionType.ExploitService, parameters={"target_host": host,"target_service": service,"source_host": src_host,}))
     # Data Scans
     for host in state.controlled_hosts:
-        valid_actions.add(Action(ActionType.FindData, params={"target_host": host, "source_host": host}))
+        if not is_fw_blocked(state, src_host,host):
+            valid_actions.add(Action(ActionType.FindData, parameters={"target_host": host, "source_host": host}))
 
     # Data Exfiltration
     for src_host, data_list in state.known_data.items():
         for data in data_list:
             for trg_host in state.controlled_hosts:
                 if trg_host != src_host:
-                    valid_actions.add(Action(ActionType.ExfiltrateData, params={"target_host": trg_host, "source_host": src_host, "data": data}))
+                    if not is_fw_blocked(state, src_host,trg_host):
+                        valid_actions.add(Action(ActionType.ExfiltrateData, parameters={"target_host": trg_host, "source_host": src_host, "data": data}))
     
-    # BlockIP
-    for src_host in state.controlled_hosts:
-        for target_host in state.controlled_hosts:
-            for blocked_ip in state.known_hosts:
-                valid_actions.add(Action(ActionType.BlockIP, {"target_host":target_host, "source_host":src_host, "blocked_host":blocked_ip}))
-
-                    
+    if include_blocks:
+        # BlockIP
+        if include_blocks:
+            for src_host in state.controlled_hosts:
+                for target_host in state.controlled_hosts:
+                    if not is_fw_blocked(state, src_host,target_host):
+                        for blocked_ip in state.known_hosts:
+                            valid_actions.add(Action(ActionType.BlockIP, {"target_host":target_host, "source_host":src_host, "blocked_host":blocked_ip}))
     return list(valid_actions)    
 
 def state_as_ordered_string(state:GameState)->str:
@@ -97,6 +110,9 @@ def state_as_ordered_string(state:GameState)->str:
     ret += "},data:{"
     for host in sorted(state.known_data.keys()):
         ret += f"{host}:[{','.join([str(x) for x in sorted(state.known_data[host])])}]"
+    ret += "},blocks:{"
+    for host in sorted(state.known_blocks.keys()):
+        ret += f"{host}:[{','.join([str(x) for x in sorted(state.known_blocks[host])])}]"
     ret += "}"
     return ret   
 
@@ -480,7 +496,7 @@ def convert_concepts_to_actions(action, concept_mapping, logger):
             if src_host_concept == host_concept:
                 new_src_host = concept_mapping['controlled_hosts'][host_concept]
 
-        action = Action(ActionType.ExploitService, params={"target_host": new_target_host, "target_service": new_target_service, "source_host": new_src_host})
+        action = Action(ActionType.ExploitService, parameters={"target_host": new_target_host, "target_service": new_target_service, "source_host": new_src_host})
 
     elif action._type == ActionType.ExfiltrateData:
         # parameters = {
@@ -508,7 +524,7 @@ def convert_concepts_to_actions(action, concept_mapping, logger):
         data_concept = action.parameters['data']
         new_data = data_concept
 
-        action = Action(ActionType.ExfiltrateData, params={"target_host": new_target_host, "source_host": new_src_host, "data": new_data})
+        action = Action(ActionType.ExfiltrateData, parameters={"target_host": new_target_host, "source_host": new_src_host, "data": new_data})
 
     elif action._type == ActionType.FindData:
         # parameters = {
@@ -529,7 +545,7 @@ def convert_concepts_to_actions(action, concept_mapping, logger):
             if src_host_concept == host_concept:
                 new_src_host = concept_mapping['controlled_hosts'][host_concept]
         
-        action = Action(ActionType.FindData, params={"target_host": new_target_host, "source_host": new_src_host})
+        action = Action(ActionType.FindData, parameters={"target_host": new_target_host, "source_host": new_src_host})
 
     elif action._type == ActionType.ScanNetwork:
         target_net_concept = action.parameters['target_network']
@@ -545,7 +561,7 @@ def convert_concepts_to_actions(action, concept_mapping, logger):
         for host_concept in concept_mapping['controlled_hosts']:
             if src_host_concept == host_concept:
                 new_src_host = concept_mapping['controlled_hosts'][host_concept]
-        action = Action(ActionType.ScanNetwork, params={"source_host": new_src_host, "target_network": new_target_network} )
+        action = Action(ActionType.ScanNetwork, parameters={"source_host": new_src_host, "target_network": new_target_network} )
 
     elif action._type == ActionType.FindServices:
         # parameters = {
@@ -565,6 +581,6 @@ def convert_concepts_to_actions(action, concept_mapping, logger):
         for host_concept in concept_mapping['controlled_hosts']:
             if src_host_concept == host_concept:
                 new_src_host = concept_mapping['controlled_hosts'][host_concept]
-        action = Action(ActionType.FindServices, params={"source_host": new_src_host, "target_host": new_target_host} )
+        action = Action(ActionType.FindServices, parameters={"source_host": new_src_host, "target_host": new_target_host} )
 
     return action
\ No newline at end of file
diff --git a/agents/attackers/concepts_q_learning/conceptual_q_agent.py b/agents/attackers/concepts_q_learning/conceptual_q_agent.py
index 2980b24..1877445 100644
--- a/agents/attackers/concepts_q_learning/conceptual_q_agent.py
+++ b/agents/attackers/concepts_q_learning/conceptual_q_agent.py
@@ -2,24 +2,21 @@
 #           Arti
 #           Sebastian Garcia. sebastian.garcia@agents.fel.cvut.cz
 import sys
-from os import path, makedirs
 import numpy as np
 import random
 import pickle
 import argparse
 import logging
-# This is used so the agent can see the environment and game component
-sys.path.append(path.dirname(path.dirname(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ) ) ))))
-sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ))))
+import mlflow
+import subprocess
+from os import path, makedirs
+from AIDojoCoordinator.game_components import Action, Observation, GameState, AgentStatus
 
 # This is used so the agent can see the environment and game component
 # with the path fixed, we can import now
-from env.game_components import Action, Observation, GameState
+sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ))))
 from base_agent import BaseAgent
 from agent_utils import generate_valid_actions, state_as_ordered_string, convert_concepts_to_actions, convert_ips_to_concepts
-import mlflow
-import subprocess
-
 
 class QAgent(BaseAgent):
 
@@ -383,15 +380,15 @@ def play_game(self, observation_ip, episode_num, testing=False):
                                 test_end = test_observation.end
                                 test_info = test_observation.info
 
-                                if test_info and test_info['end_reason'] == 'blocked':
+                                if test_info and test_info['end_reason'] == AgentStatus.Fail:
                                     test_detected +=1
                                     test_num_detected_steps += [num_steps]
                                     test_num_detected_returns += [reward]
-                                elif test_info and test_info['end_reason'] == 'goal_reached':
+                                elif test_info and test_info['end_reason'] == AgentStatus.Success:
                                     test_wins += 1
                                     test_num_win_steps += [num_steps]
                                     test_num_win_returns += [reward]
-                                elif test_info and test_info['end_reason'] == 'max_steps':
+                                elif test_info and test_info['end_reason'] == AgentStatus.TimeoutReached:
                                     test_max_steps += 1
                                     test_num_max_steps_steps += [num_steps]
                                     test_num_max_steps_returns += [reward]
diff --git a/agents/attackers/double_q_learning/double_q_agent.py b/agents/attackers/double_q_learning/double_q_agent.py
index 287d7c2..c8f4b17 100644
--- a/agents/attackers/double_q_learning/double_q_agent.py
+++ b/agents/attackers/double_q_learning/double_q_agent.py
@@ -14,8 +14,8 @@
 import logging
 from torch.utils.tensorboard import SummaryWriter
 import time
-from env.worlds.network_security_game import NetworkSecurityEnvironment
-from env.game_components import Action, Observation, GameState
+from AIDojoCoordinator.worlds.network_security_game import NetworkSecurityEnvironment
+from AIDojoCoordinator.game_components import Action, Observation, GameState
 
 class DoubleQAgent:
 
diff --git a/agents/attackers/gnn_reinforce/gnn_REINFORCE_agent.py b/agents/attackers/gnn_reinforce/gnn_REINFORCE_agent.py
index c7e9abb..3140530 100644
--- a/agents/attackers/gnn_reinforce/gnn_REINFORCE_agent.py
+++ b/agents/attackers/gnn_reinforce/gnn_REINFORCE_agent.py
@@ -18,7 +18,7 @@
 sys.path.append(path.dirname(path.dirname(path.dirname(path.dirname( path.dirname( path.abspath(__file__) ) ) ))))
 
 #with the path fixed, we can import now
-from env.worlds.network_security_game import NetworkSecurityEnvironment
+from AIDojoCoordinator.worlds.network_security_game import NetworkSecurityEnvironment
 
 os.environ['CUDA_VISIBLE_DEVICES'] = '-1'
 os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
diff --git a/agents/attackers/interactive_tui/assistant.py b/agents/attackers/interactive_tui/assistant.py
index 0ce98dd..fe11f32 100644
--- a/agents/attackers/interactive_tui/assistant.py
+++ b/agents/attackers/interactive_tui/assistant.py
@@ -10,13 +10,8 @@
 import jinja2
 from tenacity import retry, stop_after_attempt
 
-sys.path.append(
-    path.dirname(path.dirname(path.dirname(path.dirname(path.dirname(path.abspath(__file__))))))
-)
-from env.game_components import (
-    ActionType,
-    Observation,
-)
+
+from AIDojoCoordinator.game_components import ActionType, Observation
 
 sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__)))))
 
diff --git a/agents/attackers/interactive_tui/interactive_tui.py b/agents/attackers/interactive_tui/interactive_tui.py
index 94ee988..fead246 100644
--- a/agents/attackers/interactive_tui/interactive_tui.py
+++ b/agents/attackers/interactive_tui/interactive_tui.py
@@ -1,29 +1,20 @@
 #
 # Author:  Maria Rigaki - maria.rigaki@aic.fel.cvut.cz
-#
-from textual.app import App, ComposeResult, Widget
-from textual.widgets import Tree, Button, RichLog, Select, Input
-from textual.containers import Vertical, VerticalScroll, Horizontal
-from textual.validation import Function
-from textual import on
-from textual.reactive import reactive
 
 import sys
-from os import path
 import os
 import logging
 import ipaddress
 import argparse
 import asyncio
-
+from textual.app import App, ComposeResult, Widget
+from textual.widgets import Tree, Button, RichLog, Select, Input
+from textual.containers import Vertical, VerticalScroll, Horizontal
+from textual.validation import Function
+from textual import on
+from textual.reactive import reactive
 from assistant import LLMAssistant
-
-# This is used so the agent can see the environment and game components
-sys.path.append(
-    path.dirname(path.dirname(path.dirname(path.dirname(path.dirname(path.abspath(__file__))))))
-)
-from env.game_components import Network, IP
-from env.game_components import ActionType, Action, GameState, Observation
+from AIDojoCoordinator.game_components import Network, IP, ActionType, Action, GameState, Observation
 
 # This is used so the agent can see the BaseAgent
 sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
@@ -558,7 +549,7 @@ def generate_action(self, state: GameState) -> Action:
                         self.network_input[:-3], mask=int(self.network_input[-2:])
                     ),
                 }
-                action = Action(action_type=self.next_action, params=parameters)
+                action = Action(action_type=self.next_action, parameters=parameters)
             else:
                 self.notify("Please provide valid inputs", severity="error")
         elif self.next_action in [ActionType.FindServices, ActionType.FindData]:
@@ -567,7 +558,7 @@ def generate_action(self, state: GameState) -> Action:
                     "source_host": IP(self.src_host_input),
                     "target_host": IP(self.target_host_input),
                 }
-                action = Action(action_type=self.next_action, params=parameters)
+                action = Action(action_type=self.next_action, parameters=parameters)
             else:
                 self.notify("Please provide valid inputs", severity="error")
         elif self.next_action == ActionType.ExploitService:
@@ -582,7 +573,7 @@ def generate_action(self, state: GameState) -> Action:
                                     "target_service": service,
                                 }
                                 action = Action(
-                                    action_type=self.next_action, params=parameters
+                                    action_type=self.next_action, parameters=parameters
                                 )
                                 break
             else:
@@ -600,7 +591,7 @@ def generate_action(self, state: GameState) -> Action:
                                     "data": datum,
                                 }
                                 action = Action(
-                                    action_type=self.next_action, params=parameters
+                                    action_type=self.next_action, parameters=parameters
                                 )
                             else:
                                 parameters = self.data_input
diff --git a/agents/attackers/llm/llm_agent-2.py b/agents/attackers/llm/llm_agent-2.py
index 64bf6ff..a9def62 100644
--- a/agents/attackers/llm/llm_agent-2.py
+++ b/agents/attackers/llm/llm_agent-2.py
@@ -8,8 +8,8 @@
 sys.path.append(path.dirname(path.dirname(path.dirname(path.dirname( path.dirname( path.abspath(__file__) ) ) ))))
 
 
-from env.worlds.network_security_game import NetworkSecurityEnvironment
-from env.game_components import ActionType, Action, IP, Data, Network, Service
+from AIDojoCoordinator.worlds.network_security_game import NetworkSecurityEnvironment
+from AIDojoCoordinator.game_components import ActionType, Action, IP, Data, Network, Service
 
 import openai
 from tenacity import retry, stop_after_attempt
diff --git a/agents/attackers/llm/llm_agent-3.py b/agents/attackers/llm/llm_agent-3.py
index 09c8e0b..c1f997c 100644
--- a/agents/attackers/llm/llm_agent-3.py
+++ b/agents/attackers/llm/llm_agent-3.py
@@ -7,8 +7,8 @@
 sys.path.append(path.dirname(path.dirname(path.dirname(path.dirname( path.dirname( path.abspath(__file__) ) ) ))))
 
 
-from env.worlds.network_security_game import NetworkSecurityEnvironment
-from env.game_components import ActionType, Action, IP, Data, Network, Service
+from AIDojoCoordinator.worlds.network_security_game import NetworkSecurityEnvironment
+from AIDojoCoordinator.game_components import ActionType, Action, IP, Data, Network, Service
 
 import openai
 from tenacity import retry, stop_after_attempt
diff --git a/agents/attackers/llm/llm_agent.py b/agents/attackers/llm/llm_agent.py
index 36a366e..f98be5f 100644
--- a/agents/attackers/llm/llm_agent.py
+++ b/agents/attackers/llm/llm_agent.py
@@ -7,8 +7,8 @@
 sys.path.append(path.dirname(path.dirname(path.dirname(path.dirname( path.dirname( path.abspath(__file__) ) ) ))))
 
 
-from env.worlds.network_security_game import NetworkSecurityEnvironment
-from env.game_components import ActionType, Action, IP, Data, Network, Service
+from AIDojoCoordinator.worlds.network_security_game import NetworkSecurityEnvironment
+from AIDojoCoordinator.game_components import ActionType, Action, IP, Data, Network, Service
 
 import openai
 from tenacity import retry, stop_after_attempt
diff --git a/agents/attackers/llm_embed/llm_embed.py b/agents/attackers/llm_embed/llm_embed.py
index be57cc2..694819c 100644
--- a/agents/attackers/llm_embed/llm_embed.py
+++ b/agents/attackers/llm_embed/llm_embed.py
@@ -11,8 +11,8 @@
 from os import path
 sys.path.append(path.dirname(path.dirname(path.dirname(path.dirname( path.dirname( path.abspath(__file__) ) ) ))))
 
-from env.worlds.network_security_game import NetworkSecurityEnvironment
-from env.game_components import Action, ActionType
+from AIDojoCoordinator.worlds.network_security_game import NetworkSecurityEnvironment
+from AIDojoCoordinator.game_components import Action, ActionType
 
 import numpy as np
 import torch
diff --git a/agents/attackers/llm_embed_dqn/llm_embed_dqn.py b/agents/attackers/llm_embed_dqn/llm_embed_dqn.py
index 12b8184..b795439 100644
--- a/agents/attackers/llm_embed_dqn/llm_embed_dqn.py
+++ b/agents/attackers/llm_embed_dqn/llm_embed_dqn.py
@@ -13,8 +13,8 @@
 from os import path
 sys.path.append(path.dirname(path.dirname(path.dirname(path.dirname( path.dirname( path.abspath(__file__) ) ) ))))
 
-from env.worlds.network_security_game import NetworkSecurityEnvironment
-from env.game_components import Action, ActionType
+from AIDojoCoordinator.worlds.network_security_game import NetworkSecurityEnvironment
+from AIDojoCoordinator.game_components import Action, ActionType
 from sentence_transformers import SentenceTransformer
 
 import numpy as np
diff --git a/agents/attackers/llm_qa/llm_agent_qa.py b/agents/attackers/llm_qa/llm_agent_qa.py
index 862bcbf..015c6ba 100644
--- a/agents/attackers/llm_qa/llm_agent_qa.py
+++ b/agents/attackers/llm_qa/llm_agent_qa.py
@@ -5,13 +5,27 @@
 """
 import logging
 import argparse
-from llm_action_planner import LLMActionPlanner
 import logging
 import numpy as np
 import pandas as pd
 import mlflow
+import sys
+from llm_action_planner import LLMActionPlanner
+from os import path
+
+mlflow.set_tracking_uri("http://147.32.83.60")
+mlflow.set_experiment("LLM_QA")
+
+from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig
+
+sys.path.append(
+    path.dirname(path.dirname(path.dirname(path.dirname(path.abspath(__file__)))))
+)
+
+from AIDojoCoordinator.game_components import ActionType, Action, IP, Data, Network, Service
 
-from env.game_components import ActionType
+# This is used so the agent can see the BaseAgent
+sys.path.append(path.dirname(path.dirname(path.abspath(__file__))))
 from base_agent import BaseAgent
 
 mlflow.set_tracking_uri("http://147.32.83.60")
diff --git a/agents/attackers/q_learning/q_agent.py b/agents/attackers/q_learning/q_agent.py
index ae74cc1..3a18660 100644
--- a/agents/attackers/q_learning/q_agent.py
+++ b/agents/attackers/q_learning/q_agent.py
@@ -7,19 +7,15 @@
 import pickle
 import argparse
 import logging
+import mlflow
+import subprocess
 
 from os import path, makedirs
-# This is used so the agent can see the environment and game component
-sys.path.append(path.dirname(path.dirname(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ) ) ))))
-sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ))))
-
-# This is used so the agent can see the environment and game component
 # with the path fixed, we can import now
-from env.game_components import Action, Observation, GameState
+from AIDojoCoordinator.game_components import Action, Observation, GameState, AgentStatus
+sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ))))
 from base_agent import BaseAgent
 from agent_utils import generate_valid_actions, state_as_ordered_string
-import mlflow
-import subprocess
 
 
 class QAgent(BaseAgent):
@@ -288,15 +284,15 @@ def play_game(self, observation, episode_num, testing=False):
                     end = observation.end
                     info = observation.info
 
-                    if observation.info and observation.info['end_reason'] == 'blocked':
+                    if observation.info and observation.info['end_reason'] == AgentStatus.Fail:
                         detected +=1
                         num_detected_steps += [num_steps]
                         num_detected_returns += [reward]
-                    elif observation.info and observation.info['end_reason'] == 'goal_reached':
+                    elif observation.info and observation.info['end_reason'] == AgentStatus.Success:
                         wins += 1
                         num_win_steps += [num_steps]
                         num_win_returns += [reward]
-                    elif observation.info and observation.info['end_reason'] == 'max_steps':
+                    elif observation.info and observation.info['end_reason'] == AgentStatus.TimeoutReached:
                         max_steps += 1
                         num_max_steps_steps += [num_steps]
                         num_max_steps_returns += [reward]
diff --git a/agents/attackers/random/random_agent.py b/agents/attackers/random/random_agent.py
index f307852..0e22159 100644
--- a/agents/attackers/random/random_agent.py
+++ b/agents/attackers/random/random_agent.py
@@ -2,18 +2,16 @@
 # This agents just randomnly picks actions. No learning
 import sys
 import logging
-import os
-from random import choice
 import argparse
 import numpy as np
 import mlflow
+from os import path, makedirs
+from random import choice
+from AIDojoCoordinator.game_components import Action, Observation, AgentStatus
 
-# This is used so the agent can see the environment and game components
-sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__) ) ) ))))
-sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__) ))))
+sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ))))
 
 # with the path fixed, we can import now
-from env.game_components import Action, Observation
 from base_agent import BaseAgent
 from agent_utils import generate_valid_actions
 from datetime import datetime
@@ -131,15 +129,15 @@ def select_action(self, observation:Observation)->Action:
                 end = observation.end
                 info = observation.info
 
-                if observation.info and observation.info['end_reason'] == 'blocked':
+                if observation.info and observation.info['end_reason'] == AgentStatus.Fail:
                     detected +=1
                     num_detected_steps += [num_steps]
                     num_detected_returns += [reward]
-                elif observation.info and observation.info['end_reason'] == 'goal_reached':
+                elif observation.info and observation.info['end_reason'] == AgentStatus.Success:
                     wins += 1
                     num_win_steps += [num_steps]
                     num_win_returns += [reward]
-                elif observation.info and observation.info['end_reason'] == 'max_steps':
+                elif observation.info and observation.info['end_reason'] == AgentStatus.TimeoutReached:
                     max_steps += 1
                     num_max_steps_steps += [num_steps]
                     num_max_steps_returns += [reward]
diff --git a/agents/attackers/sarsa/sarsa_agent.py b/agents/attackers/sarsa/sarsa_agent.py
index 8246408..382f378 100644
--- a/agents/attackers/sarsa/sarsa_agent.py
+++ b/agents/attackers/sarsa/sarsa_agent.py
@@ -8,16 +8,10 @@
 import argparse
 import logging
 
-# This is used so the agent can see the environment and game components
-sys.path.append(
-    path.dirname(path.dirname(path.dirname(path.dirname(path.dirname(path.abspath(__file__))))))
-)
+from AIDojoCoordinator.game_components import Action, GameState
 # importing agent utils and base agent
 sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ))))
-
-# This is used so the agent can see the environment and game component
 # with the path fixed, we can import now
-from env.game_components import Action, GameState
 from base_agent import BaseAgent
 from agent_utils import generate_valid_actions, state_as_ordered_string
 
@@ -104,7 +98,9 @@ def play_game(self, num_episodes=1, testing=False):
                     testing_returns = []
                     for _ in range(args.eval_for):
                         testing_returns.append(np.sum(self.play_episode(testing=True)))
-                    self._logger.info(f"Eval after {episode} episodes: ={np.mean(testing_returns)}±{np.std(testing_returns)}")                
+                    self._logger.info(f"Eval after {episode} episodes: ={np.mean(testing_returns)}±{np.std(testing_returns)}")
+                if episode % args.store_models_every == 0 and episode != 0:
+                    self.store_q_table(f'sarsa_agent_marl.experiment{args.experiment_id}-episodes-{episode}.pickle')           
         returns = []
         for _ in range(args.eval_for):
             returns.append(np.sum(self.play_episode(testing=True)))
@@ -116,14 +112,17 @@ def play_game(self, num_episodes=1, testing=False):
     parser = argparse.ArgumentParser()
     parser.add_argument("--host", help="Host where the game server is", default="127.0.0.1", action='store', required=False)
     parser.add_argument("--port", help="Port where the game server is", default=9000, type=int, action='store', required=False)
-    parser.add_argument("--episodes", help="Sets number of testing episodes", default=5, type=int)
-    parser.add_argument("--eval_each", help="Sets periodic evaluation during training", default=250, type=int)
-    parser.add_argument("--eval_for", help="Sets length of periodic evaluation", default=200, type=int)
+    parser.add_argument("--episodes", help="Sets number of testing episodes", default=25000, type=int)
+    parser.add_argument("--eval_each", help="Sets periodic evaluation during training", default=5000, type=int)
+    parser.add_argument("--eval_for", help="Sets length of periodic evaluation", default=5000, type=int)
     parser.add_argument("--epsilon", help="Sets epsilon for exploration", default=0.2, type=float)
     parser.add_argument("--gamma", help="Sets gamma for Q learing", default=0.9, type=float)
-    parser.add_argument("--alpha", help="Sets alpha for learning rate", default=0.8, type=float)
+    parser.add_argument("--alpha", help="Sets alpha for learning rate", default=0.1, type=float)
     parser.add_argument("--logdir", help="Folder to store logs", default=path.join(path.dirname(path.abspath(__file__)), "logs"))
     parser.add_argument("--test_only", help="Only run testing", default=False, action='store_true')
+    parser.add_argument("--experiment_id", help="Id of the experiment to record into Mlflow.", default='sarsa_006_coordinatorV3', type=str)
+    parser.add_argument("--store_models_every", help="Store a model to disk every these number of episodes.", default=2000, type=int)
+    parser.add_argument("--previous_model", help="Store a model to disk every these number of episodes.", type=str)
     args = parser.parse_args()
 
     if not path.exists(args.logdir):
@@ -134,8 +133,9 @@ def play_game(self, num_episodes=1, testing=False):
     agent = SARSAAgent(args.host, args.port, alpha=args.alpha, gamma=args.gamma, epsilon=args.epsilon)
 
     if args.test_only:
-        agent.load_q_table("./sarsa_agent_marl.pickle")
+        agent.load_q_table(args.previous_model)
         agent.play_game(args.episodes, testing=True)       
     else:
         agent.play_game(args.episodes, testing=False)
-        agent.store_q_table("./sarsa_agent_marl.pickle")
\ No newline at end of file
+        agent.store_q_table("./sarsa_agent_marl.pickle")
+
diff --git a/agents/attackers/scripted_attacker/scripted_attacker.py b/agents/attackers/scripted_attacker/scripted_attacker.py
new file mode 100644
index 0000000..f1811a1
--- /dev/null
+++ b/agents/attackers/scripted_attacker/scripted_attacker.py
@@ -0,0 +1,85 @@
+import sys
+import argparse
+import logging
+from os import path
+import time
+
+from AIDojoCoordinator.game_components import Action, ActionType, IP, Network
+sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ))))
+#with the path fixed, we can import now
+from base_agent import BaseAgent
+
+def winning_strat(host, port, delay=10):
+    agent1 = BaseAgent(host, port, role="Attacker")
+    obs1 = agent1.register()
+    time.sleep(delay)
+    print(obs1)
+    print("----------------------------")
+    # # network scan
+    obs1 = agent1.make_step(Action(
+        ActionType.ScanNetwork,
+        parameters={
+            "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
+            "target_network":Network("192.168.1.0", 24) 
+            }
+        )
+    )
+    print(obs1)
+    time.sleep(delay)
+    print("----------------------------")
+    obs1 = agent1.make_step(Action(
+        ActionType.FindServices,
+        parameters={
+            "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
+            "target_host": IP("192.168.1.2")
+            }
+        )
+    )
+    print(obs1)
+    print("----------------------------")
+    time.sleep(delay)
+    obs1 = agent1.make_step(Action(
+        ActionType.ExploitService,
+        parameters={
+            "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
+            "target_host": IP("192.168.1.2"),
+            "target_service": list(obs1.state.known_services[IP("192.168.1.2")])[0]
+            }
+        )
+    )
+    print(obs1)
+    print("----------------------------")
+    time.sleep(delay)
+    obs1 = agent1.make_step(Action(
+        ActionType.FindData,
+        parameters={
+            "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
+            "target_host": IP("192.168.1.2"),
+            }
+        )
+    )
+    print(obs1)
+    print("----------------------------")
+    time.sleep(delay)
+    obs1 = agent1.make_step(Action(
+        ActionType.ExfiltrateData,
+        parameters={
+            "source_host": IP("192.168.1.2"),
+            "target_host": IP("213.47.23.195"),
+            "data":list(obs1.state.known_data[IP("192.168.1.2")])[0]
+            }
+        )
+    )
+    print(obs1)
+    print("----------------------------")
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--host", help="Host where the game server is", default="127.0.0.1", action='store', required=False)
+    parser.add_argument("--port", help="Port where the game server is", default=9000, type=int, action='store', required=False)
+    parser.add_argument("--delay", help="Delay between actions (in seconds)", default=10, type=int, action='store', required=False)
+    
+    args = parser.parse_args()
+    log_filename = path.dirname(path.abspath(__file__)) + '/base_agent.log'
+    logging.basicConfig(filename="scripted_attacker.log", filemode='w', format='%(asctime)s %(name)s %(levelname)s %(message)s',  datefmt='%Y-%m-%d %H:%M:%S', level=logging.DEBUG)
+    winning_strat(args.host, args.port, args.delay)
\ No newline at end of file
diff --git a/agents/base_agent.py b/agents/base_agent.py
index cdbdb4e..64668cc 100644
--- a/agents/base_agent.py
+++ b/agents/base_agent.py
@@ -1,19 +1,16 @@
 # Author: Ondrej Lukas, ondrej.lukas@aic.cvut.cz
 # Basic agent class that is to be extended in each agent classes
 import sys
-import argparse
 import logging
 
 from os import path
 import socket
 import json
-from abc import ABC
-
+from abc import ABC 
 
 # This is used so the agent can see the environment and game components
 sys.path.append(path.dirname(path.dirname( path.dirname( path.abspath(__file__) ) ) ))
-from env.game_components import Action, GameState, Observation, ActionType, GameStatus,AgentInfo, IP, Network, Data
-
+from AIDojoCoordinator.game_components import Action, GameState, Observation, ActionType, GameStatus,AgentInfo
 
 class BaseAgent(ABC):
     """
@@ -101,7 +98,7 @@ def _receive_data(socket)->tuple:
             return GameStatus.from_string(status), observation, message
         
         if isinstance(data, Action):
-            data = data.as_json()
+            data = data.to_json()
         else:
             raise ValueError("Incorrect data type! Data should be ONLY of type Action")
         
@@ -116,7 +113,7 @@ def register(self)->Observation:
         try:
             self._logger.info(f'Registering agent as {self.role}')
             status, observation_dict, message = self.communicate(Action(ActionType.JoinGame,
-                                                                         params={"agent_info":AgentInfo(self.__class__.__name__,self.role)}))
+                                                                         parameters={"agent_info":AgentInfo(self.__class__.__name__,self.role)}))
             self._logger.info(f'\tRegistering agent as {status, observation_dict, message}')
             if status is GameStatus.CREATED:
                 self._logger.info(f"\tRegistration successful! {message}")
@@ -132,294 +129,10 @@ def request_game_reset(self)->Observation:
         Method for requesting restart of the game.
         """
         self._logger.debug("Requesting game reset")
-        status, observation_dict, message = self.communicate(Action(ActionType.ResetGame))
+        status, observation_dict, message = self.communicate(Action(ActionType.ResetGame, {}))
         if status:
             self._logger.debug('\tReset successful')
             return Observation(GameState.from_dict(observation_dict["state"]), observation_dict["reward"], observation_dict["end"], message)
         else:
             self._logger.error(f'\rReset failed! (status: {status}, msg:{message}')
             return None
-
-
-if __name__ == '__main__':
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--host", help="Host where the game server is", default="127.0.0.1", action='store', required=False)
-    parser.add_argument("--port", help="Port where the game server is", default=9000, type=int, action='store', required=False)
-    
-    args = parser.parse_args()
-    log_filename = path.dirname(path.abspath(__file__)) + '/base_agent.log'
-    logging.basicConfig(filename=log_filename, filemode='w', format='%(asctime)s %(name)s %(levelname)s %(message)s',  datefmt='%Y-%m-%d %H:%M:%S', level=logging.DEBUG)
-   
-    # # #######################################
-    # # ATTACKER X ATTACKER 
-    # # #######################################
-    
-    # #register both agents
-    # agent1 = BaseAgent(args.host, args.port, role="Attacker")
-    # obs1 = agent1.register()
-    # agent2 = BaseAgent(args.host, args.port, role="Attacker")
-    # obs2 = agent2.register()
-
-    # # network scans - both agents
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.ScanNetwork,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-    #         "target_network":Network("192.168.1.0", 24)
-    #         }
-    #     )
-    # )
-    # obs2 = agent2.make_step(Action(
-    #     ActionType.ScanNetwork,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs2.state.controlled_hosts))[0],
-    #         "target_network":Network("192.168.1.0", 24)
-    #         }
-    #     )
-    # )
-    # # service scans - both agents
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.FindServices,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-    #         "target_host":IP("192.168.1.2")
-    #         }
-    #     )
-    # )
-    # obs2 = agent2.make_step(Action(
-    #     ActionType.FindServices,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs2.state.controlled_hosts))[0],
-    #         "target_host":IP("192.168.1.2")
-    #         }
-    #     )
-    # )
-    # # agent 1 exploit
-    # host, services  = [(k,v) for k,v in obs1.state.known_services.items()][0]
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.ExploitService,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-    #         "target_host":host,
-    #         "target_service":list(services)[0]
-    #         }
-    #     )
-    # )
-    # # agent 1 find data
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.FindData,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-    #         "target_host":host,
-    #         }
-    #     )
-    # )
-
-    # # agent 1 exfiltrate data to C&C
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.ExfiltrateData,
-    #     params={
-    #         "source_host": host,
-    #         "target_host": IP("213.47.23.195"),
-    #         "data": Data("User1","DataFromServer1")
-    #         }
-    #     )
-    # )
-
-    # # agent 2 find data
-    # obs2 = agent2.make_step(Action(
-    #     ActionType.FindData,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs2.state.controlled_hosts))[0],
-    #         "target_host":host,
-    #         }
-    #     )
-    # )
-    # # agent 2 find data in C&C
-    # obs2 = agent2.make_step(Action(
-    #     ActionType.FindData,
-    #     params={
-    #         "source_host": IP("213.47.23.195"),
-    #         "target_host": IP("213.47.23.195"),
-    #         }
-    #     )
-    # )
-    # print(obs1)
-    # print(obs2)
-
-
-    ########################################
-    # ATTACKER X DEFENDER 
-    ########################################
-
-    # # register both agents
-    # agent1 = BaseAgent(args.host, args.port, role="Attacker")
-    # obs1 = agent1.register()
-    # agent2 = BaseAgent(args.host, args.port, role="Defender")
-    # obs2 = agent2.register()
-
-    # # attacker - scan net
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.ScanNetwork,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-    #         "target_network":Network("192.168.1.0", 24)
-    #         }
-    #     )
-    # )
-    
-    # # attacker - scan services
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.FindServices,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-    #         "target_host":IP("192.168.1.2")
-    #         }
-    #     )
-    # )
-    
-    # # agent 1 exploit
-    # host, services  = [(k,v) for k,v in obs1.state.known_services.items()][0]
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.ExploitService,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-    #         "target_host":host,
-    #         "target_service":list(services)[0]
-    #         }
-    #     )
-    # )
-    # # agent 1 find data
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.FindData,
-    #     params={
-    #         "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-    #         "target_host":host,
-    #         }
-    #     )
-    # )
-    # # agent 2 block connection to C&C
-    # obs2 = agent2.make_step(Action(
-    #     ActionType.BlockIP,
-    #     params={
-    #         "source_host": host,
-    #         "target_host": host,
-    #         "blocked_host": IP("213.47.23.195")
-    #         }
-    #     )
-    # )
-    # # agent 1 exfiltrate data to C&C
-    # obs1 = agent1.make_step(Action(
-    #     ActionType.ExfiltrateData,
-    #     params={
-    #         "source_host": host,
-    #         "target_host": IP("213.47.23.195"),
-    #         "data": Data("User1","DataFromServer1")
-    #         }
-    #     )
-    # )
-
-    # print(obs1)
-    # print(obs2)
-
-    # #######################################
-    # 1 ATTACKER 
-    # #######################################
-    
-    #register both agents
-    agent1 = BaseAgent(args.host, args.port, role="Attacker")
-    obs1 = agent1.register()
-    print(obs1)
-    
-    # network scans - both agents
-    obs1 = agent1.make_step(Action(
-        ActionType.ScanNetwork,
-        params={
-            "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-            "target_network":Network("192.168.1.0", 24)
-            }
-        )
-    )
-
-    # service scans - both agents
-    obs1 = agent1.make_step(Action(
-        ActionType.FindServices,
-        params={
-            "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-            "target_host":IP("192.168.1.2")
-            }
-        )
-    )
-
-    # agent 1 exploit
-    host, services  = [(k,v) for k,v in obs1.state.known_services.items()][0]
-    obs1 = agent1.make_step(Action(
-        ActionType.ExploitService,
-        params={
-            "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-            "target_host":host,
-            "target_service":list(services)[0]
-            }
-        )
-    )
-    # agent 1 find data
-    obs1 = agent1.make_step(Action(
-        ActionType.FindData,
-        params={
-            "source_host": list(filter(lambda x: x !=  IP("213.47.23.195"), obs1.state.controlled_hosts))[0],
-            "target_host":host,
-            }
-        )
-    )
-
-    # agent 1 exfiltrate data to C&C
-    obs1 = agent1.make_step(Action(
-        ActionType.ExfiltrateData,
-        params={
-            "source_host": host,
-            "target_host": IP("213.47.23.195"),
-            "data": Data("User1","DataFromServer1")
-            }
-        )
-    )
-
-    print(obs1)
-
-
-
-
-
-    obs1 = agent1.request_game_reset()
-    #obs2 = agent2.request_game_reset()
-    print("------------------")
-    print(obs1)
-    #print(obs2)
-    # #obs = agent.request_game_reset()
-    # for ip in obs.state.controlled_hosts:
-    #     if ip != IP("213.47.23.195"):
-    #         src = ip
-    #         break
-    # agent.make_step(Action(ActionType.ScanNetwork, params={"source_host":src, "target_network":Network("192.168.1.0", 24)}))
-    # obs = agent.make_step(Action(ActionType.FindServices, params={"source_host":src, "target_host":IP("192.168.1.2")}))
-    # host, services  = [(k,v) for k,v in obs.state.known_services.items()][0]
-    # service = list(services)[0]
-    # agent.make_step(Action(ActionType.ExploitService, params={"source_host":src, "target_host":host, "target_service":service}))
-    # agent.make_step(Action(action_type=ActionType.FindData, params={"source_host":host, "target_host":host}))
-    # obs = agent.make_step(Action(action_type=ActionType.ExfiltrateData, params={
-    #     "source_host":host,
-    #     "target_host":IP("213.47.23.195"),
-    #     "data": Data("User1","DataFromServer1")
-    #     }
-    # ))
-    # print(obs)
-    # # print("----------------------------")
-    # # obs = agent.request_game_reset()
-    # # print('---------------------------')
-    # # obs = agent.make_step(Action(action_type=ActionType.ExfiltrateData, params={
-    # #     "source_host":host,
-    # #     "target_host":IP("213.47.23.195"),
-    # #     "data": Data("User1","DataFromServer1")
-    # #     }
-    # # ))
-    # # print(obs)
-
diff --git a/agents/benign/random/benign_random_agent.py b/agents/benign/random/benign_random_agent.py
index e09d021..6b8cc64 100644
--- a/agents/benign/random/benign_random_agent.py
+++ b/agents/benign/random/benign_random_agent.py
@@ -8,13 +8,10 @@
 from random import uniform
 import numpy as np
 import time
-#from torch.utils.tensorboard import SummaryWriter
 
-# This is used so the agent can see the environment and game components
-sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__) ) ) )))
-sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__) )))
-# with the path fixed, we can import now
-from env.game_components import Action, Observation, ActionType
+from AIDojoCoordinator.game_components import Action, Observation, ActionType
+# importing agent utils and base agent
+sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__) ))))
 from base_agent import BaseAgent
 from agent_utils import generate_valid_actions
 
@@ -69,7 +66,7 @@ def select_action(self, observation:Observation)->Action:
         valid_actions = generate_valid_actions(observation.state)
         # filter actions based on the allowed action types
         allowed_actions = filter(lambda action: action.type in self._allowed_actions, valid_actions)
-        allowed_actions = [a for a  in allowed_actions] + [Action(ActionType.ResetGame, params={})]
+        allowed_actions = [a for a  in allowed_actions] + [Action(ActionType.ResetGame, parameters={})]
         action = choice([a for a  in allowed_actions])
         return action
 
diff --git a/agents/defenders/probabilistic/probabilistic_agent.py b/agents/defenders/probabilistic/probabilistic_agent.py
index 5f6994a..6279389 100644
--- a/agents/defenders/probabilistic/probabilistic_agent.py
+++ b/agents/defenders/probabilistic/probabilistic_agent.py
@@ -4,17 +4,19 @@
 import sys
 import logging
 import os
-from random import choice
 import argparse
-from random import uniform
 import numpy as np
 import time
-
+from random import choice
+from random import uniform
 # This is used so the agent can see the environment and game components
-sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__) ) ) )))
-sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__) )))
+sys.path.append(
+    os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))))
+)
 # with the path fixed, we can import now
-from env.game_components import Action, Observation, ActionType
+from AIDojoCoordinator.game_components import Action, Observation, ActionType
+# importing agent utils and base agent
+sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__) ))))
 from base_agent import BaseAgent
 from agent_utils import generate_valid_actions
 
@@ -67,7 +69,7 @@ def select_action(self, observation:Observation)->Action:
         valid_actions = generate_valid_actions(observation.state)
         # filter actions based on the allowed action types
         allowed_actions = filter(lambda action: action.type in self._allowed_actions, valid_actions)
-        allowed_actions = [a for a  in allowed_actions] + [Action(ActionType.ResetGame, params={})]
+        allowed_actions = [a for a  in allowed_actions] + [Action(ActionType.ResetGame, parameters={})]
         action = choice([a for a  in allowed_actions])
         return action
 
diff --git a/agents/defenders/random/random_agent.py b/agents/defenders/random/random_agent.py
index 5e7b515..30a3ea3 100644
--- a/agents/defenders/random/random_agent.py
+++ b/agents/defenders/random/random_agent.py
@@ -9,12 +9,13 @@
 import numpy as np
 import time
 import mlflow
+from os import path
 
-# This is used so the agent can see the environment and game components
-sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__) ) ) ))))
-sys.path.append(os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__) ))))
+
+# importing agent utils and base agent
+sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ))))
 # with the path fixed, we can import now
-from env.game_components import Action, Observation, ActionType
+from AIDojoCoordinator.game_components import Action, Observation, ActionType
 from base_agent import BaseAgent
 from agent_utils import generate_valid_actions
 
@@ -71,7 +72,7 @@ def select_action(self, observation:Observation)->Action:
         valid_actions = generate_valid_actions(observation.state)
         # filter actions based on the allowed action types
         allowed_actions = filter(lambda action: action.type in self._allowed_actions, valid_actions)
-        allowed_actions = [a for a  in allowed_actions] + [Action(ActionType.ResetGame, params={})]
+        allowed_actions = [a for a  in allowed_actions] + [Action(ActionType.ResetGame, parameters={})]
         action = choice([a for a  in allowed_actions])
         return action
 
diff --git a/agents/graph_agent_utils.py b/agents/graph_agent_utils.py
index 3c25881..2b8ad16 100644
--- a/agents/graph_agent_utils.py
+++ b/agents/graph_agent_utils.py
@@ -12,7 +12,7 @@
 
 sys.path.append(path.dirname(path.dirname(path.dirname(path.abspath(__file__) ))))
 #with the path fixed, we can import now
-from env.game_components import GameState, IP, Network, Data, Service
+from AIDojoCoordinator.game_components import GameState, IP, Network, Data, Service
 
 def state_as_graph(state:GameState) -> tuple:
     node_types = {
diff --git a/agents/llm_utils.py b/agents/llm_utils.py
index a5ac76b..408abd5 100644
--- a/agents/llm_utils.py
+++ b/agents/llm_utils.py
@@ -8,7 +8,7 @@
 
 sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 # with the path fixed, we can import now
-from env.game_components import (
+from AIDojoCoordinator.game_components import (
     ActionType,
     Action,
     IP,