unaiverse 0.1.11__cp311-cp311-macosx_11_0_arm64.whl

unaiverse/agent.py

@@ -0,0 +1,2090 @@
+"""
+       █████  █████ ██████   █████           █████ █████   █████ ██████████ ███████████    █████████  ██████████
+      ░░███  ░░███ ░░██████ ░░███           ░░███ ░░███   ░░███ ░░███░░░░░█░░███░░░░░███  ███░░░░░███░░███░░░░░█
+       ░███   ░███  ░███░███ ░███   ██████   ░███  ░███    ░███  ░███  █ ░  ░███    ░███ ░███    ░░░  ░███  █ ░ 
+       ░███   ░███  ░███░░███░███  ░░░░░███  ░███  ░███    ░███  ░██████    ░██████████  ░░█████████  ░██████   
+       ░███   ░███  ░███ ░░██████   ███████  ░███  ░░███   ███   ░███░░█    ░███░░░░░███  ░░░░░░░░███ ░███░░█   
+       ░███   ░███  ░███  ░░█████  ███░░███  ░███   ░░░█████░    ░███ ░   █ ░███    ░███  ███    ░███ ░███ ░   █
+       ░░████████   █████  ░░█████░░████████ █████    ░░███      ██████████ █████   █████░░█████████  ██████████
+        ░░░░░░░░   ░░░░░    ░░░░░  ░░░░░░░░ ░░░░░      ░░░      ░░░░░░░░░░ ░░░░░   ░░░░░  ░░░░░░░░░  ░░░░░░░░░░ 
+                 A Collectionless AI Project (https://collectionless.ai)
+                 Registration/Login: https://unaiverse.io
+                 Code Repositories:  https://github.com/collectionlessai/
+                 Main Developers:    Stefano Melacci (Project Leader), Christian Di Maio, Tommaso Guidi
+"""
+import copy
+import json
+import uuid
+import torch
+from unaiverse.stats import Stats
+from unaiverse.dataprops import DataProps
+from unaiverse.agent_basics import AgentBasics
+from unaiverse.streams import BufferedDataStream
+from unaiverse.networking.p2p.messages import Msg
+
+
+class Agent(AgentBasics):
+    """This class contains those basic actions that can be performed by every agent."""
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+        # Status variables (assumed to start with "_"): Agent exchanges
+        self._available = True  # It will be automatically set/changed during the agent's life
+        self._found_agents = set()  # Peer IDs discovered
+        self._valid_cmp_agents = set()  # Agents for which the last evaluation was positive
+        self._engaged_agents = set()
+        self._agents_who_completed_what_they_were_asked = set()
+        self._agents_who_were_asked = set()
+        self._eval_results = {}
+
+        # Status variables (assumed to start with "_"): Recordings
+        self._last_recorded_stream_num = 1
+        self._last_recorded_stream_dict = None
+        self._last_recording_stream_dict = None
+
+        # Status variables (assumed to start with "_"): Playlist
+        self._preferred_streams = []  # List of preferred streams
+        self._cur_preferred_stream = 0  # ID of the current preferred stream from the list
+        self._repeat = 1  # Number of repetitions of the playlist
+        
+        # Stats
+        self.stats = Stats(is_world=False)
+        self.overwrite_stats = False  # Whether to overwrite stats when receiving the next STATS_RESPONSE from the world
+
+    def remove_peer_from_agent_status_attrs(self, peer_id):
+        super().remove_peer_from_agent_status_attrs(peer_id)
+        self._available = len(self._engaged_agents) == 0
+
+    def reset_agent_status_attrs(self):
+        super().reset_agent_status_attrs()  # this sets status vars to [], {}, 0, 0., False, in function of their type
+        self._available = True
+        self._repeat = 1
+        self._last_recorded_stream_num = 1
+
+    async def set_next_action(self, agent: str | None, action: str, args: dict | None = None,
+                              ref_uuid: str | None = None):
+        """Try to tell another agent what is the next action it should run (async).
+
+        Args:
+            agent: The ID of the agent to send the action to or a valid wildcard like "<valid_cmp>" for a set of agents
+                (if None the agents in self._engaged_agents will be considered).
+            action: The name of the action to be executed by the agent.
+            args: A dictionary of arguments for the action. Defaults to None.
+            ref_uuid: An optional UUID for referencing the action. Defaults to None.
+
+        Returns:
+            True if the action was successfully sent to the target agent or to at least one of the
+            involved agents (wildcard case).
+        """
+
+        # - if "agent" is a peer ID, the involved agents will be a list with one element.
+        # - if "agent" is a known wildcard, as "<valid_cmp>", then involved agents will be self._valid_cmp_agents
+        # - if "agent" is None, then the current agent in self._engaged_agents will be returned
+        involved_agents = self.__involved_agents(agent)
+        if len(involved_agents) == 0:
+            return False
+
+        at_least_one_completed = False
+        _, private_peer_id = self.get_peer_ids()
+        for _peer_id in involved_agents:
+            ret = await self._node_conn.send(_peer_id, channel_trail=None,
+                                             content={"action_name": action, "args": args, "uuid": ref_uuid},
+                                             content_type=Msg.ACTION_REQUEST)
+            at_least_one_completed = at_least_one_completed or ret
+            self.deb(f"[set_next_action] {self._node_name} sent action: {action}, with args: {args}, "
+                     f"and result of sending is {ret}")
+        return at_least_one_completed
+
+    async def set_engaged_partner(self, agent: str):
+        """Virtually forces the engagement with a single agent, clearing all existing engagements (async).
+
+        Returns:
+            True all the times.
+        """
+        self._engaged_agents.clear()
+        self._engaged_agents.add(agent)
+        return True
+
+    async def send_engagement(self):
+        """Offer engagement to the agents whose identifiers are in self._found_agents (async).
+
+        Returns:
+            True if engagement requests were successfully sent to at least one found agent, False otherwise.
+        """
+        at_least_one_sent = False
+
+        if len(self._found_agents) > 0:
+            self.out(f"Sending engagement request to {', '.join([x for x in self._found_agents])}")
+        my_role_str = self._node_profile.get_dynamic_profile()['connections']['role']
+        for found_agent in self._found_agents:
+            if await self.set_next_action(found_agent, action="get_engagement",
+                                          args={"sender_role": my_role_str}):
+                at_least_one_sent = True
+            else:
+                self.err(f"Unable to send engagement to {found_agent}")
+
+        return at_least_one_sent
+
+    async def get_engagement(self, acceptable_role: str | None = None, sender_role: str | None = None,
+                             _requester: str | None = None):
+        """Receive engagement from another agent whose authority is in the specified range (async).
+
+        Args:
+            acceptable_role: The role that the sender must have for engagement to be accepted. Defaults to None.
+            sender_role: The role of the agent sending the engagement request. Defaults to None.
+            _requester: The ID of the agent requesting engagement (automatically set by the action calling routine)
+
+        Returns:
+            True if the engagement was successfully received and confirmed, False otherwise.
+        """
+        self.out(f"Getting engagement from {_requester}, whose role is {sender_role} (looking for {acceptable_role})")
+        if _requester not in self.world_agents and _requester not in self.world_masters:
+            self.err(f"Unknown agent: {_requester}")
+            return False
+
+        if sender_role is None:
+            self.err(f"Unknown role of {_requester}")
+            return False
+
+        # Confirming
+        if self._available:
+            acceptable_role_int = self.ROLE_STR_TO_BITS[acceptable_role]
+            if "~" not in acceptable_role:
+                sender_role_int = (self.ROLE_STR_TO_BITS[sender_role] >> 2) << 2
+            else:
+                sender_role_int = self.ROLE_STR_TO_BITS[sender_role]
+
+            if acceptable_role_int == sender_role_int:
+                if await self.set_next_action(_requester, "got_engagement"):
+                    self._engaged_agents.add(_requester)
+
+                    # Marking this agent as not available since it engaged with another one
+                    self._available = False
+                    return True
+                else:
+                    self.err(f"Unable to confirm engagement to {_requester}")
+                    return False
+            else:
+                self.err(f"Cannot engage to {_requester}")
+                return False
+        else:
+            self.err(f"Cannot engage to {_requester}")
+            return False
+
+    async def got_engagement(self, _requester: str | None = None):
+        """Confirm an engagement (async).
+
+        Args:
+            _requester: The ID of the agent confirming the engagement (automatically set by the action calling routine).
+
+        Returns:
+            True if the engagement was successfully confirmed, False otherwise.
+        """
+        self.out(f"Confirming engagement with {_requester}")
+        if _requester in self._found_agents:
+            self._engaged_agents.add(_requester)
+
+            # Marking this agent as not available since it engaged with another one
+            self._available = False
+
+            # Removing the engaged agent from the list of found agents, to avoid sending him another engagement request
+            self._found_agents.discard(_requester)
+            return True
+        else:
+            self.err(f"Unable to confirm engagement with {_requester}")
+            return False
+
+    async def send_disengagement(self, send_disconnection_too: bool = False):
+        """Ask for disengagement (async).
+
+        Args:
+            send_disconnection_too: Whether to send a disconnect-suggestion together with the disengagement.
+
+        Returns:
+            True if disengagement requests were successfully sent to at least one engaged agent, False otherwise.
+        """
+        at_least_one_sent = False
+
+        if len(self._engaged_agents) > 0:
+            self.out(f"Sending disengagement request to {', '.join([x for x in self._engaged_agents])}")
+        for agent in self._engaged_agents:
+            if await self.set_next_action(agent, action="get_disengagement",
+                                          args={"disconnect_too": send_disconnection_too}):
+                at_least_one_sent = True
+            else:
+                self.err(f"Unable to send disengagement to {agent}")
+
+        return at_least_one_sent
+
+    async def get_disengagement(self, disconnect_too: bool = False, _requester: str | None = None):
+        """Get a disengagement request from an agent (async).
+
+        Args:
+            disconnect_too: Whether to disconnect the agent who sent the disengagement.
+            _requester: The ID of the agent requesting disengagement. Defaults to None.
+
+        Returns:
+            True if the disengagement request was successfully processed, False otherwise.
+        """
+        self.out(f"Getting a disengagement request from {_requester}")
+        if _requester not in self.world_agents and _requester not in self.world_masters:
+            self.err(f"Unknown agent: {_requester}")
+            return False
+
+        if _requester not in self._engaged_agents:
+            self.err(f"Not previously engaged to {_requester}")
+            return False
+
+        if disconnect_too:
+            await self._node_purge_fcn(_requester)
+
+        self._engaged_agents.discard(_requester)  # Remove if present
+
+        # Marking this agent as available if not engaged to any agent
+        self._available = len(self._engaged_agents) == 0
+        return True
+
+    async def disengage_all(self):
+        """Disengage all the previously engaged agents (async).
+
+        Returns:
+            True if the disengagement procedure was successfully executed, False otherwise.
+        """
+        self.out(f"Disengaging all agents")
+        self._engaged_agents = set()
+
+        # Marking this agent as available
+        self._available = True
+        return True
+
+    async def disconnect_by_role(self, role: str | list[str]):
+        """Disconnects from all agents that match a specified role (async).
+        It finds the agents and calls the node's purge function on each.
+
+        Args:
+            role: A string or list of strings representing the role(s) of agents to disconnect from.
+
+        Returns:
+            Always True.
+        """
+        self.out(f"Disconnecting agents with role: {role}")
+        if await self.find_agents(role):
+            found_agents = copy.deepcopy(self._found_agents)
+            for agent in found_agents:
+                await self._node_purge_fcn(agent)  # This will also call remove_agent, that will call remove_streams
+        return True
+
+    async def disconnected(self, agent: str | None = None, delay: float = -1.):
+        """Checks if a specific set of agents (by ID or wildcard) are no longer connected to the agent.
+        It returns False if any of the specified agents are still connected (async).
+
+        Args:
+            agent: The ID of the agent or a wildcard to check.
+            delay: The time (seconds) to be spent in the current state before actually considering this action.
+
+        Returns:
+            True if all involved agents are disconnected, False otherwise.
+
+        """
+        assert delay is not None, "Missing basic action information"
+
+        # - if "agent" is a peer ID, the involved agents will be a list with one element.
+        # - if "agent" is a known wildcard, as "<valid_cmp>", then involved agents will be self._valid_cmp_agents
+        # - if "agent" is None, then the current agent in self._engaged_agents will be returned
+        involved_agents = self.__involved_agents(agent)
+        if len(involved_agents) == 0:
+            return False
+
+        self.out(f"Checking if all these agents are not connected to me anymore: {involved_agents}")
+        all_disconnected = True
+        for agent in involved_agents:
+            if agent in self.world_agents or agent in self.public_agents or agent in self._node_agents_waiting\
+                    or self._node_conn.is_connected(agent):
+                all_disconnected = False
+                break
+        return all_disconnected
+
+    async def received_some_asked_data(self, processing_fcn: str | None = None):
+        """Checks if any of the agents that were previously asked for data (e.g., via `ask_gen`) have sent a stream
+        sample back. Optionally, it can process the received data with a specified function (async).
+
+        Args:
+            processing_fcn: The name of a function to process the received data.
+
+        Returns:
+            True if at least one data sample was received, False otherwise.
+        """
+        _processing_fcn = None
+        if processing_fcn is not None:
+            if hasattr(self, processing_fcn):
+                _processing_fcn = getattr(self, processing_fcn)
+                if not callable(_processing_fcn):
+                    _processing_fcn = None
+            if _processing_fcn is None:
+                self.err(f"Processing function not found: {processing_fcn}")
+
+        got_something = False
+        for agent in self._agents_who_were_asked:
+            net_hash_to_stream_dict = self.find_streams(agent, "processor")
+            for stream_dict in net_hash_to_stream_dict.values():
+                for stream_obj in stream_dict.values():
+                    if not stream_obj.props.is_public():
+                        data = stream_obj.get("received_some_asked_data")
+                        data_tag = stream_obj.get_tag()
+
+                        if data is not None:
+                            if _processing_fcn is None:
+                                return True
+                            else:
+                                got_something = True
+                                _processing_fcn(agent, stream_obj.props, data, data_tag)
+        return got_something
+
+    async def nop(self, message: str | None = None, delay: float = -1.):
+        """Do nothing (async).
+
+        Args:
+            message: An optional message to print. Defaults to None.
+            delay: The time (seconds) to be spent in the current state before actually considering this action.
+
+        Returns:
+            Always True.
+        """
+        assert delay is not None, "Missing basic action information"
+        if message is not None:
+            self.out(message)
+        return True
+
+    async def wait_for_actions(self, agent: str, from_state: str, to_state: str, wait: bool):
+        """Lock or unlock every action between a pair of states in the state machine of a target agent (async).
+
+        Args:
+            agent: The ID of the agent to send the action locking request to, or a valid wildcard like "<valid_cmp>"
+                for a set of agents (if None the agents in self._engaged_agents will be considered).
+            from_state: The starting state of the actions to be locked/unlocked.
+            to_state: The ending state of the actions to be locked/unlocked.
+            wait: A boolean indicating whether to wait for the actions to complete (wait == !ready).
+
+        Returns:
+            True if the request was successfully sent to at least one involved agent, False otherwise.
+        """
+
+        # - if "agent" is a peer ID, the involved agents will be a list with one element.
+        # - if "agent" is a known wildcard, as "<valid_cmp>", then involved agents will be self._valid_cmp_agents
+        # - if "agent" is None, then the current agent in self._engaged_agents will be returned
+        involved_agents = self.__involved_agents(agent)
+        if len(involved_agents) == 0:
+            return False
+
+        at_least_one_completed = False
+        for _agent in involved_agents:
+            self.out(f"Telling {_agent} to alter his HSM {from_state} -> {to_state} (wait: {wait}) "
+                     f"by calling method 'wait_for_actions' on it")
+            ret = await self._node_conn.send(_agent, channel_trail=None,
+                                             content={'method': 'wait_for_actions',
+                                                      'args': (from_state, to_state, wait)},
+                                             content_type=Msg.HSM)
+            at_least_one_completed = at_least_one_completed or ret
+        return at_least_one_completed
+
+    async def ask_gen(self, agent: str | None = None, u_hashes: list[str] | None = None,
+                      samples: int = 100, time: float = -1., timeout: float = -1., ask_uuid: str | None = None,
+                      ignore_uuid: bool = False):
+        """Asking for generation.
+
+        Args:
+            agent: The ID of the agent to ask for generation, or a valid wildcard like "<valid_cmp>"
+                for a set of agents (if None the agents in self._engaged_agents will be considered).
+            u_hashes: A list of input stream hashes for generation. Defaults to None.
+            samples: The number of samples to generate. Defaults to 100.
+            time: The time duration for generation. Defaults to -1.
+            timeout: The timeout for the generation request. Defaults to -1.
+            ask_uuid: Specify the UUID of the action (if None - default -, it is randomly generated).
+            ignore_uuid: Force a None UUID instead of generating a random one.
+
+        Returns:
+            True if the generation request was successfully sent to at least one involved agent, False otherwise.
+        """
+        assert samples is not None and time is not None and timeout is not None, "Missing basic action information"
+
+        # - if "agent" is a peer ID, the involved agents will be a list with one element.
+        # - if "agent" is a known wildcard, as "<valid_cmp>", then involved agents will be self._valid_cmp_agents
+        # - if "agent" is None, then the current agent in self._engaged_agents will be returned
+        involved_agents = self.__involved_agents(agent)
+        self.deb(f"[ask_gen] Involved_agents: {involved_agents}")
+
+        if len(involved_agents) == 0:
+            self.deb(f"[ask_gen] No involved agents, action ask_gen returns False")
+            return False
+
+        # Create a copy of the input hashes, normalizing them in the appropriate way
+        u_hashes_copy: list[str | None] = [None] * len(u_hashes)
+        for i in range(len(u_hashes_copy)):
+            if u_hashes_copy[i] == "<playlist>":
+
+                # From <playlist> to the current element of the playlist
+                u_hashes_copy[i] = self._preferred_streams[self._cur_preferred_stream]
+            else:
+
+                # From a user specified hash to a net hash (e.g., peer_id:name_or_group to peer_id::ps:name_or_group)
+                u_hashes_copy[i] = self.user_stream_hash_to_net_hash(u_hashes[i])
+
+        # Generate a new UUID for this request
+        ref_uuid = uuid.uuid4().hex[0:8] if ask_uuid is None else ask_uuid
+        if ignore_uuid:
+            ref_uuid = None
+
+        # If the input streams are all owned by this agent, discard UUID
+        all_owned = True
+        for i in range(len(u_hashes_copy)):
+            if u_hashes_copy[i] not in self.owned_streams:
+                all_owned = False
+                break
+        if not all_owned:
+            ref_uuid = None
+
+        for i in range(len(u_hashes_copy)):
+
+            # If there are our own streams involved, and they are buffered, let's plan to restart them when we will
+            # start sending them through the net: moreover, let's set the local stream UUID appropriately to
+            # the generated UUID
+            if u_hashes_copy[i] in self.owned_streams:
+                stream_dict = self.known_streams[u_hashes_copy[i]]
+                for stream_name, stream_obj in stream_dict.items():
+
+                    # Plan to restart buffered streams
+                    if isinstance(stream_obj, BufferedDataStream):
+                        stream_obj.plan_restart_before_next_get(requested_by="send_stream_samples")
+
+                    # Activate the stream (if it was off)
+                    stream_obj.enable()
+
+                    # Set UUID to the generated one
+                    stream_obj.set_uuid(ref_uuid=ref_uuid, expected=False)
+                    stream_obj.set_uuid(ref_uuid=None, expected=True)
+
+        self.deb(f"[ask_gen] Input streams u_hashes: {u_hashes_copy}")
+
+        self.out(f"Asking {', '.join(involved_agents)} to generate signal given {u_hashes_copy} (ref_uuid: {ref_uuid})")
+        self._agents_who_completed_what_they_were_asked = set()
+        self._agents_who_were_asked = set()
+        correctly_asked = []
+        for peer_id in involved_agents:
+            ret = await self.__ask_gen_or_learn(for_what="gen", agent=peer_id,
+                                                u_hashes=u_hashes_copy,
+                                                yhat_hashes=None,
+                                                samples=samples, time=time, timeout=timeout, ref_uuid=ref_uuid)
+            self.deb(f"[ask_gen] Asking {peer_id} returned {ret}")
+            if ret:
+                correctly_asked.append(peer_id)
+
+        # Preparing the buffered stream where to store data, if needed
+        if len(correctly_asked) > 0:
+
+            # Saving
+            self.last_ref_uuid = ref_uuid
+
+            # For each agent that we involve in this request....
+            for peer_id in correctly_asked:
+
+                # Finding the streams generated by the processor of the agent we asked to generate
+                processor_streams = self.find_streams(peer_id, name_or_group="processor")
+
+                # For each stream generated by the processor of the agent we asked to generate...
+                for net_hash, stream_dict in processor_streams.items():
+
+                    # Set the appropriate UUID to the one we created in this method
+                    for stream in stream_dict.values():
+                        stream.set_uuid(None, expected=False)
+                        stream.set_uuid(ref_uuid, expected=True)  # Setting the "expected" one
+
+        self.deb(f"[ask_gen] Overall, the action ask_gen will return {len(correctly_asked) > 0}")
+        return len(correctly_asked) > 0
+
+    async def do_gen(self, u_hashes: list[str] | None = None,
+                     samples: int = 100, time: float = -1., timeout: float = -1.,
+                     _requester: str | list | None = None, _request_time: float = -1., _request_uuid: str | None = None,
+                     _completed: bool = False) -> bool:
+        """Generate a signal (async).
+
+        Args:
+            u_hashes: A list of input stream hashes for generation. Defaults to None.
+            samples: The number of samples to generate. Defaults to 100.
+            time: The max time duration for whole generation process. Defaults to -1.
+            timeout: The timeout for generation attempts: if calling the generate action fails for more than "timeout"
+            seconds, it is declared as complete. Defaults to -1.
+            _requester: The ID of the agent who requested generation (automatically set by the action calling routine).
+            _request_time: The time the generation was requested (automatically set by the action calling routine).
+            _request_uuid: The UUID of the generation request (automatically set by the action calling routine).
+            _completed: A boolean indicating if the generation is already completed (automatically set by the action
+                calling routine). This will tell that it is time to run a final procedure.
+
+        Returns:
+            True if the signal generation was successful, False otherwise.
+        """
+        assert samples is not None and time is not None and timeout is not None, "Missing basic action information"
+
+        self.deb(f"[do_gen] Samples: {samples}, time: {time}, timeout: {timeout}, "
+                 f"requester: {_requester}, request_time: {_request_time}, request_uuid: {_request_uuid}, "
+                 f"completed: {_completed}")
+
+        if _requester is not None:
+            if isinstance(_requester, list):
+                for _r in _requester:
+                    if self.behaving_in_world():
+                        if _r not in self.world_agents and _requester not in self.world_masters:
+                            self.err(f"Unknown agent: {_r} in list {_requester} (fully skipping generation)")
+                            return False
+                    else:
+                        if _r not in self.public_agents:
+                            self.err(f"Unknown agent: {_r} in list {_requester} (fully skipping generation)")
+                            return False
+            else:
+                if self.behaving_in_world():
+                    if _requester not in self.world_agents and _requester not in self.world_masters:
+                        self.err(f"Unknown agent: {_requester} (fully skipping generation)")
+                        return False
+                else:
+                    if _requester not in self.public_agents:
+                        self.err(f"Unknown agent: {_requester} (fully skipping generation)")
+                        return False
+
+        # Check what is the step ID of the multistep action
+        k = self.get_action_step()
+
+        # In the first step of this action, we change the UUID of the local stream associated to the input data we will
+        # use to handle this action, setting expectations to avoid handling tags of old data
+        if k == 0:
+
+            # Warning: we are not normalizing the hashes, we should do it if this action is called directly
+            if u_hashes is not None:
+                for net_hash in u_hashes:
+                    if net_hash in self.known_streams:
+                        for stream_name, stream_obj in self.known_streams[net_hash].items():
+
+                            # If the data arrived before this action, then the UUID is already set, and here there is
+                            # no need to do anything; if the data has not yet arrived (common case) ...
+                            if stream_obj.get_uuid(expected=False) != _request_uuid:
+                                stream_obj.set_uuid(None, expected=False)  # Clearing UUID
+                                stream_obj.set_uuid(_request_uuid, expected=True)  # Setting expectations
+                    else:
+                        self.out(f"Unknown stream mentioned in u_hashes: {net_hash}")
+                        return False
+
+        if not _completed:
+            self.out(f"Generating signal")
+            ret = self.__process_streams(u_hashes=u_hashes, yhat_hashes=None, learn=False,
+                                         recipient=_requester, ref_uuid=_request_uuid)
+            if not ret:
+                self.out(f"Generating signal failed")
+            else:
+                if not self.is_multi_steps_action():
+                    self.out(f"Completing signal generation (degenerate single-step case of a multi-step action")
+                    ret = await self.__complete_do(do_what="gen", peer_id_who_asked=_requester, all_hashes=u_hashes,
+                                                   send_back_confirmation=False)
+                    if not ret:
+                        self.out(f"Completing signal generation failed")
+            return ret
+        else:
+            self.out(f"Completing signal generation")
+            ret = await self.__complete_do(do_what="gen", peer_id_who_asked=_requester, all_hashes=u_hashes)
+            if not ret:
+                self.out(f"Completing signal generation failed")
+            return ret
+
+    async def done_gen(self, _requester: str | None = None):
+        """This is a way to get back the confirmation of a completed generation (async).
+
+        Args:
+            _requester: The ID of the agent who completed the generation. Defaults to None.
+
+        Returns:
+            True if the generation confirmation was successfully handled by this agent, False is something went wrong.
+        """
+        self.out(f"Agent {_requester} finished generation")
+
+        # Searching for the processor-streams of the agent who generated data
+        processor_streams = self.find_streams(_requester, name_or_group="processor")
+        if processor_streams is None or len(processor_streams) == 0:
+            self.err("Unexpected confirmation of finished generation")
+            return False
+
+        # Remembering that the agent that invoked this action is the one who generated the data, and what he generated
+        # could be used in future action (for example, in evaluation processes)
+        self._agents_who_completed_what_they_were_asked.add(_requester)
+
+        # Clearing the UUID of the local streams associated to the agent who generated
+        for net_hash, stream_dict in processor_streams.items():
+            for stream_obj in stream_dict.values():
+                stream_obj.set_uuid(None, expected=False)
+                stream_obj.set_uuid(None, expected=True)
+
+        # If one or more of my streams where used as arguments of the generation request I did (ask_gen), then their
+        # UUID must be cleared...we clear them all
+        for net_hash, stream_dict in self.owned_streams.items():
+            for stream_obj in stream_dict.values():
+                if stream_obj.props.is_public() != self.behaving_in_world():
+                    stream_obj.set_uuid(None, expected=False)
+                    stream_obj.set_uuid(None, expected=True)
+        return True
+
+    async def ask_learn(self, agent: str | None = None,
+                        u_hashes: list[str] | None = None, yhat_hashes: list[str] | None = None,
+                        samples: int = 100, time: float = -1., timeout: float = -1., ask_uuid: str | None = None,
+                        ignore_uuid: str | None = None):
+        """Asking for learning to generate (async).
+
+        Args:
+            agent: The ID of the agent to ask for generation, or a valid wildcard like "<valid_cmp>"
+                for a set of agents (if None the agents in self._engaged_agents will be considered).
+            u_hashes: A list of input stream hashes for inference. Defaults to None.
+            yhat_hashes: A list of target stream hashes to be used for loss computation. Defaults to None.
+            samples: The number of samples to learn from. Defaults to 100.
+            time: The time duration for generation. Defaults to -1.
+            timeout: The timeout for the generation request. Defaults to -1.
+            ask_uuid: Specify the action UUID (default = None, i.e., it is automatically generated).
+            ignore_uuid: If Trie, the UUID is fully ignored (i.e, forced to None).
+
+        Returns:
+            True if the learning request was successfully sent to at least one involved agent, False otherwise.
+        """
+        assert samples is not None and time is not None and timeout is not None, "Missing basic action information"
+
+        # - if "agent" is a peer ID, the involved agents will be a list with one element.
+        # - if "agent" is a known wildcard, as "<valid_cmp>", then involved agents will be self._valid_cmp_agents
+        # - if "agent" is None, then the current agent in self._engaged_agents will be returned
+        involved_agents = self.__involved_agents(agent)
+        self.deb(f"[ask_learn] Involved agents: {involved_agents}")
+
+        if len(involved_agents) == 0:
+            self.deb(f"[ask_learn] No involved agents, action will return False")
+            return False
+
+        # Create a copy of the input hashes, normalizing them in the appropriate way
+        u_hashes_copy = [x for x in u_hashes]
+        for i in range(len(u_hashes_copy)):
+            if u_hashes_copy[i] == "<playlist>":
+
+                # From <playlist> to the current element of the playlist
+                u_hashes_copy[i] = self._preferred_streams[self._cur_preferred_stream]
+            else:
+
+                # From a user specified hash to a net hash (e.g., peer_id:name_or_group to peer_id::ps:name_or_group)
+                u_hashes_copy[i] = self.user_stream_hash_to_net_hash(u_hashes_copy[i])
+
+        # Create a copy of the target hashes, normalizing them in the appropriate way
+        yhat_hashes_copy = [x for x in yhat_hashes]
+        for i in range(len(yhat_hashes_copy)):
+            if yhat_hashes_copy[i] == "<playlist>":
+
+                # From <playlist> to the current element of the playlist
+                yhat_hashes_copy[i] = self._preferred_streams[self._cur_preferred_stream]
+            else:
+
+                # From a user specified hash to a net hash (e.g., peer_id:name_or_group to peer_id::ps:name_or_group)
+                yhat_hashes_copy[i] = self.user_stream_hash_to_net_hash(yhat_hashes_copy[i])
+
+        # Generate a new UUID for this request
+        ref_uuid = uuid.uuid4().hex[0:8] if ask_uuid is None else ask_uuid
+        if ignore_uuid:
+            ref_uuid = None
+
+        # If the input streams are all owned by this agent, discard UUID
+        all_owned = True
+        for i in range(len(u_hashes_copy)):
+            if u_hashes_copy[i] not in self.owned_streams:
+                all_owned = False
+                break
+        if all_owned:
+            for i in range(len(yhat_hashes_copy)):
+                if yhat_hashes_copy[i] not in self.owned_streams:
+                    all_owned = False
+                    break
+        if not all_owned:
+            ref_uuid = None
+
+        for i in range(len(u_hashes_copy)):
+
+            # If there are our own streams involved, and they are buffered, let's plan to restart them when we will
+            # start sending them through the net: moreover, let's set the local stream UUID appropriately to
+            # the generated UUID
+            if u_hashes_copy[i] in self.owned_streams:
+                stream_dict = self.known_streams[u_hashes_copy[i]]
+                for stream_name, stream_obj in stream_dict.items():
+
+                    # Plan to restart buffered streams
+                    if isinstance(stream_obj, BufferedDataStream):
+                        stream_obj.plan_restart_before_next_get(requested_by="send_stream_samples")
+
+                    # Activate the stream (if it was off)
+                    stream_obj.enable()
+
+                    # Set UUID to the generated one
+                    stream_obj.set_uuid(ref_uuid=ref_uuid, expected=False)
+                    stream_obj.set_uuid(ref_uuid=None, expected=True)
+
+        for i in range(len(yhat_hashes_copy)):
+
+            # If there are our own streams involved, and they are buffered, let's plan to restart them when we will
+            # start sending them through the net: moreover, let's set the local stream UUID appropriately to
+            # the generated UUID
+            if yhat_hashes_copy[i] in self.owned_streams:
+                stream_dict = self.known_streams[yhat_hashes_copy[i]]
+                for stream_name, stream_obj in stream_dict.items():
+
+                    # Plan to restart buffered streams
+                    if isinstance(stream_obj, BufferedDataStream):
+                        stream_obj.plan_restart_before_next_get(requested_by="send_stream_samples")
+
+                    # Activate the stream (if it was off)
+                    stream_obj.enable()
+
+                    # Set UUID to the generated one
+                    stream_obj.set_uuid(ref_uuid=ref_uuid, expected=False)
+                    stream_obj.set_uuid(ref_uuid=None, expected=True)
+
+        self.out(f"Asking {', '.join(involved_agents)} to learn to generate signal {yhat_hashes_copy}, "
+                 f"given {u_hashes_copy} (ref_uuid: {ref_uuid})")
+        self._agents_who_completed_what_they_were_asked = set()
+        self._agents_who_were_asked = set()
+        correctly_asked = []
+        for peer_id in involved_agents:
+            ret = await self.__ask_gen_or_learn(for_what="learn", agent=peer_id,
+                                                u_hashes=u_hashes_copy,
+                                                yhat_hashes=yhat_hashes_copy,
+                                                samples=samples, time=time, timeout=timeout, ref_uuid=ref_uuid)
+            self.deb(f"[ask_learn] Asking {peer_id} returned {ret}")
+            if ret:
+                correctly_asked.append(peer_id)
+
+        # Preparing the buffered stream where to store data, if needed
+        if len(correctly_asked) > 0:
+
+            # Saving
+            self.last_ref_uuid = ref_uuid
+
+            # For each agent that we involve in this request....
+            for peer_id in correctly_asked:
+
+                # Finding the streams generated by the processor of the agent we asked to generate
+                processor_streams = self.find_streams(peer_id, name_or_group="processor")
+
+                # For each stream generated by the processor of the agent we asked to generate...
+                for net_hash, stream_dict in processor_streams.items():
+
+                    # Set the appropriate UUID to the one we created in this method
+                    for stream in stream_dict.values():
+                        stream.set_uuid(None, expected=False)
+                        stream.set_uuid(ref_uuid, expected=True)  # Setting the "expected" one
+
+        self.deb(f"[ask_learn] Overall the action ask_learn will return {len(correctly_asked) > 0}")
+        return len(correctly_asked) > 0
+
+    async def do_learn(self, yhat_hashes: list[str] | None = None, u_hashes: list[str] | None = None,
+                       samples: int = 100, time: float = -1., timeout: float = -1.,
+                       _requester: str | None = None, _request_time: float = -1., _request_uuid: str | None = None,
+                       _completed: bool = False) -> bool:
+        """Learn to generate a signal (async).
+
+        Args:
+            yhat_hashes: A list of target stream hashes to be used for loss computation. Defaults to None.
+            u_hashes: A list of input stream hashes for inference. Defaults to None.
+            samples: The number of samples to learn from. Defaults to 100.
+            time: The max time duration of the learning procedure. Defaults to -1.
+            timeout: The timeout for learning attempts: if calling the learning action fails for more than "timeout"
+            seconds, it is declared as complete. Defaults to -1.
+            _requester: The ID of the agent who requested learning (automatically set by the action calling routine).
+            _request_time: The time learning was requested (automatically set by the action calling routine).
+            _request_uuid: The UUID of the learning request (automatically set by the action calling routine).
+            _completed: A boolean indicating if the learning is already completed (automatically set by the action
+                calling routine). This will tell that it is time to run a final procedure.
+
+        Returns:
+            True if the signal generation was successful, False otherwise.
+        """
+        assert samples is not None and time is not None and timeout is not None, "Missing basic action information"
+
+        self.deb(f"[do_learn] samples: {samples}, time: {time}, timeout: {timeout}, "
+                 f"requester: {_requester}, request_time: {_request_time}, request_uuid: {_request_uuid} "
+                 f"completed: {_completed}")
+
+        if _requester not in self.world_agents and _requester not in self.world_masters:
+            self.err(f"Unknown agent: {_requester}")
+            return False
+
+        # Check what is the step ID of the multistep action
+        k = self.get_action_step()
+
+        # In the first step of this action, we change the UUID of the local stream associated to the input data we will
+        # use to handle this action, setting expectations to avoid handling tags of old data
+        if k == 0:
+
+            # Warning: we are not normalizing the hashes, we should do it if this action is called directly
+            if u_hashes is not None:
+                for net_hash in u_hashes:
+                    if net_hash in self.known_streams:
+                        for stream_obj in self.known_streams[net_hash].values():
+
+                            # If the data arrived before this action, then the UUID is already set, and here there is
+                            # no need to do anything; if the data has not yet arrived (common case) ...
+                            if stream_obj.get_uuid(expected=False) != _request_uuid:
+                                stream_obj.set_uuid(None, expected=False)  # Clearing UUID
+                                stream_obj.set_uuid(_request_uuid, expected=True)  # Setting expectations
+
+            # Warning: we are not normalizing the hashes, we should do it if this action is called directly
+            if yhat_hashes is not None:
+                for net_hash in yhat_hashes:
+                    if net_hash in self.known_streams:
+                        for stream_obj in self.known_streams[net_hash].values():
+                            if stream_obj.get_uuid(expected=False) != _request_uuid:
+                                stream_obj.set_uuid(None, expected=False)  # Clearing UUID
+                                stream_obj.set_uuid(_request_uuid, expected=True)  # Setting expectations
+
+        if not _completed:
+            self.out(f"Learning to generate signal {yhat_hashes}")
+            ret = self.__process_streams(u_hashes=u_hashes, yhat_hashes=yhat_hashes, learn=True,
+                                         recipient=_requester, ref_uuid=_request_uuid)
+            if not ret:
+                self.out(f"Learning to generate signal {yhat_hashes} failed")
+            return ret
+        else:
+            self.out(f"Completing learning to generate signal {yhat_hashes}")
+            all_hashes = (u_hashes if u_hashes is not None else []) + (yhat_hashes if yhat_hashes is not None else [])
+            ret = await self.__complete_do(do_what="learn", peer_id_who_asked=_requester, all_hashes=all_hashes)
+            if not ret:
+                self.out(f"Completing learning to generate signal {yhat_hashes} failed")
+            return ret
+
+    async def done_learn(self, _requester: str | None = None):
+        """This is a way to get back the confirmation of a completed learning procedure (async).
+
+        Args:
+            _requester: The ID of the agent who completed the learning procedure. Defaults to None.
+
+        Returns:
+            True if the learning-complete confirmation was successfully handled by this agent, False otherwise.
+        """
+        self.out(f"Agent {_requester} finished learning")
+        self._agents_who_completed_what_they_were_asked.add(_requester)
+
+        # Searching for the processor-streams of the agent who generated the (inference) data
+        processor_streams = self.find_streams(_requester, name_or_group="processor")
+        if processor_streams is None or len(processor_streams) == 0:
+            self.err("Unexpected confirmation of finished learning")
+            return False
+
+        # Warning: differently from the case of done_gen, we are not considering the streams generated by the
+        # learning agents as something we could use for evaluation (this might be changed in the future)
+
+        # Clearing the UUID of the local streams associated to the agent who learned
+        for net_hash, stream_dict in processor_streams.items():
+            for stream_obj in stream_dict.values():
+                stream_obj.set_uuid(None, expected=False)
+                stream_obj.set_uuid(None, expected=True)
+
+        # If one or more of my streams where used as arguments of the learning request I did (ask_learn), then their
+        # UUID must be cleared...we clear them all
+        for net_hash, stream_dict in self.owned_streams.items():
+            for stream_obj in stream_dict.values():
+                if stream_obj.props.is_public() != self.behaving_in_world():
+                    stream_obj.set_uuid(None, expected=False)
+                    stream_obj.set_uuid(None, expected=True)
+        return True
+
+    async def all_asked_finished(self):
+        """Checks if all agents that were previously asked to perform a task (e.g., generate or learn) have sent a
+        completion confirmation. It compares the set of agents asked with the set of agents that have completed
+        the task (async).
+
+        Returns:
+            True if all agents are done, False otherwise.
+        """
+        return self._agents_who_were_asked == self._agents_who_completed_what_they_were_asked
+
+    async def all_engagements_completed(self):
+        """Checks if all engagement requests that were sent have been confirmed. It returns True if there are no agents
+        remaining in the `_found_agents` list, implying all have been engaged with or discarded (async).
+
+        Returns:
+            True if all engagements are complete, False otherwise.
+
+        """
+        return len(self._found_agents) == 0
+
+    async def agents_are_waiting(self):
+        """Checks if there are any agents who have connected but have not yet been fully processed or added to the
+        agent's known lists. This indicates that new agents are waiting to be managed (async).
+
+        Returns:
+            True if there are waiting agents, False otherwise.
+        """
+        self.out(f"Current set of {len(self._node_agents_waiting)} connected peer IDs non managed yet: "
+                 f"{self._node_agents_waiting}")
+        for found_agent in self._found_agents:
+            if found_agent in self._node_agents_waiting:
+                return True
+        return False
+
+    async def ask_subscribe(self, agent: str | None = None,
+                            stream_hashes: list[str] | None = None, unsubscribe: bool = False):
+        """Requests a remote agent or a group of agents to subscribe to or unsubscribe from a list of specified PubSub
+        streams. It normalizes the stream hashes and sends an action request containing the stream properties (async).
+
+        Args:
+            agent: The target agent's ID or a wildcard.
+            stream_hashes: A list of streams to subscribe to or unsubscribe from.
+            unsubscribe: A boolean to indicate if it's an unsubscription request.
+
+        Returns:
+            True if the request was sent to at least one agent, False otherwise.
+        """
+
+        # - if "agent" is a peer ID, the involved agents will be a list with one element.
+        # - if "agent" is a known wildcard, as "<valid_cmp>", then involved agents will be self._valid_cmp_agents
+        # - if "agent" is None, then the current agent in self._engaged_agents will be returned
+        involved_agents = self.__involved_agents(agent)
+        self.deb(f"[ask_subscribe] Involved_agents: {involved_agents}")
+
+        if len(involved_agents) == 0:
+            self.deb(f"[ask_subscribe] No involved agents, action ask_gen returns False")
+            return False
+
+        # Create a copy of the stream hashes, normalizing them in the appropriate way
+        stream_hashes_copy: list[str | None] = [None] * len(stream_hashes)
+        for i in range(len(stream_hashes_copy)):
+            if stream_hashes_copy[i] == "<playlist>":
+
+                # From <playlist> to the current element of the playlist
+                stream_hashes_copy[i] = self._preferred_streams[self._cur_preferred_stream]
+            else:
+
+                # From a user specified hash to a net hash (e.g., peer_id:name_or_group to peer_id::ps:name_or_group)
+                stream_hashes_copy[i] = self.user_stream_hash_to_net_hash(stream_hashes[i])
+
+        # Getting properties
+        stream_owners = []
+        stream_props = []
+        for i in range(len(stream_hashes_copy)):
+            stream_dict = self.known_streams[stream_hashes_copy[i]]
+            peer_id = DataProps.peer_id_from_net_hash(stream_hashes_copy[i])
+            for name, stream_obj in stream_dict.items():
+                stream_owners.append(peer_id)
+                stream_props.append(json.dumps(stream_obj.props.to_dict()))
+
+        what = "subscribe to" if not unsubscribe else "unsubscribe from "
+        self.out(f"Asking {', '.join(involved_agents)} to {what} {stream_hashes}")
+        self._agents_who_completed_what_they_were_asked = set()
+        self._agents_who_were_asked = set()
+        correctly_asked = []
+        for agent in involved_agents:
+            if await self.set_next_action(agent, action="do_subscribe", args={"stream_owners": stream_owners,
+                                                                              "stream_props": stream_props,
+                                                                              "unsubscribe": unsubscribe}):
+                self._agents_who_were_asked.add(agent)
+                ret = True
+            else:
+                what = "subscribe" if not unsubscribe else "unsubscribe"
+                self.err(f"Unable to ask {agent} to {what}")
+                ret = False
+            self.deb(f"[ask_subscribe] Asking {agent} returned {ret}")
+            if ret:
+                correctly_asked.append(agent)
+
+        self.deb(f"[ask_subscribe] Overall, the action ask_subscribe (unsubscribe: {unsubscribe})"
+                 f" will return {len(correctly_asked) > 0}")
+        return len(correctly_asked) > 0
+
+    async def do_subscribe(self, stream_owners: list[str] | None = None, stream_props: list[str] | None = None,
+                           unsubscribe: bool = False,
+                           _requester: str | list | None = None, _request_time: float = -1.):
+        """Executes a subscription or unsubscription request received from another agent. It processes the stream
+        properties, adds or removes the streams from the agent's known streams, and handles the underlying PubSub topic
+        subscriptions (async).
+
+        Args:
+            stream_owners: A list of peer IDs who own the streams.
+            stream_props: A list of JSON-serialized stream properties.
+            unsubscribe: A boolean to indicate unsubscription.
+            _requester: The ID of the requesting agent.
+            _request_time: The time the request was made.
+
+        Returns:
+            True if the action is successful, False otherwise.
+        """
+        self.deb(f"[do_subscribe] unsubscribe: {unsubscribe}, "
+                 f"stream_owners: {stream_owners}, stream_props: ... ({len(stream_props)} props)")
+
+        if _requester is not None:
+            if isinstance(_requester, list):
+                for _r in _requester:
+                    if self.behaving_in_world():
+                        if _r not in self.world_agents and _requester not in self.world_masters:
+                            self.err(f"Unknown agent: {_r} in list {_requester} (fully skipping do_subscribe)")
+                            return False
+                    else:
+                        if _r not in self.public_agents:
+                            self.err(f"Unknown agent: {_r} in list {_requester} (fully skipping do_subscribe)")
+                            return False
+            else:
+                if self.behaving_in_world():
+                    if _requester not in self.world_agents and _requester not in self.world_masters:
+                        self.err(f"Unknown agent: {_requester} (fully skipping do_subscribe)")
+                        return False
+                else:
+                    if _requester not in self.public_agents:
+                        self.err(f"Unknown agent: {_requester} (fully skipping do_subscribe)")
+                        return False
+        else:
+            self.err("Unknown requester (None)")
+            return False
+
+        # Building properties
+        props_dicts = []
+        props_objs = []
+        for i in range(len(stream_props)):
+            p_dict = json.loads(stream_props[i])
+            props = DataProps.from_dict(p_dict)
+            if props.is_pubsub():
+                props_dicts.append(p_dict)
+                props_objs.append(props)
+            else:
+                self.err(f"Expecting a pubsub stream, got a stream named {props.get_name()} "
+                         f"(group is {props.get_group()}), which is not pubsub")
+                return False
+
+        # Adding new streams and subscribing (if compatible with our processor)
+        for stream_owner, prop_dict, prop_obj in zip(stream_owners, props_dicts, props_objs):
+            if not unsubscribe:
+                if not (await self.add_compatible_streams(peer_id=stream_owner, streams_in_profile=[prop_dict],
+                                                          buffered=False, public=False)):
+                    self.out(f"Unable to add a pubsub stream ({prop_obj.get_name()}) from agent {stream_owner}: "
+                             f"no compatible streams were found")
+            else:
+                if not (await self.remove_streams(peer_id=stream_owner, name=prop_obj.get_name())):
+                    self.out(f"Unable to unsubscribe from pubsub stream ({prop_obj.get_name()}) "
+                             f"of agent {stream_owner}")
+        return True
+
+    async def done_subscribe(self, unsubscribe: bool = False, _requester: str | None = None):
+        """Handles the confirmation that a subscription or unsubscription request has been completed by another agent.
+        It adds the requester to the set of agents that have completed their asked tasks (async).
+
+        Args:
+            unsubscribe: A boolean indicating if it was an unsubscription.
+            _requester: The ID of the agent who completed the task.
+
+        Returns:
+            Always True.
+        """
+        what = "subscribing" if unsubscribe else "unsubscribing"
+        self.out(f"Agent {_requester} finished {what}")
+
+        # Remembering that the agent that invoked this action is the one who actually subscribed
+        self._agents_who_completed_what_they_were_asked.add(_requester)
+        return True
+
+    async def record(self, net_hash: str, samples: int = 100, time: float = -1., timeout: float = -1.):
+        """Records data from a specified stream into a new, owned `BufferedDataStream`. This is a multistep action
+        that captures a sequence of samples over time and then adds the new recorded stream to the agent's profile
+        (async).
+
+        Args:
+            net_hash: The hash of the stream to record.
+            samples: The number of samples to record.
+            time: The time duration for recording.
+            timeout: The timeout for each recording attempt.
+
+        Returns:
+            True if a sample was successfully recorded, False otherwise.
+        """
+        assert samples is not None and time is not None and timeout is not None, "Missing basic action information"
+
+        k = self.get_action_step()
+
+        self.out(f"Recording stream {net_hash}")
+
+        if k == 0:
+
+            # Getting stream(s)
+            _net_hash = self.user_stream_hash_to_net_hash(net_hash)  # In case of ambiguity, it yields the first one
+            if _net_hash is None:
+                self.err(f"Unknown stream {net_hash}")
+                return False
+            else:
+                net_hash = _net_hash
+
+            stream_src_dict = self.known_streams[net_hash]
+
+            # Creating the new recorded stream (same props of the recorded one, just owned now)
+            stream_dest_dict = {}
+            for name, stream_obj in stream_src_dict.items():
+                props = stream_obj.props.clone()
+                props.set_group("recorded" + str(self._last_recorded_stream_num))
+                stream_dest_dict[name] = BufferedDataStream(props=props, clock=self._node_clock)
+            self._last_recorded_stream_dict = stream_dest_dict
+            self._last_recording_stream_dict = stream_src_dict
+
+        else:
+
+            # Retrieving the stream(s)
+            stream_dest_dict = self._last_recorded_stream_dict
+            stream_src_dict = self._last_recording_stream_dict
+
+        # Recording
+        for name, stream_obj in stream_src_dict.items():
+            x = stream_obj.get(requested_by="record")
+            if x is None:
+                self.deb("[record] data sample missing, returning False")
+                return False
+            else:
+                self.deb(f"[record] data_tag: {stream_obj.get_tag()}, data_uuid: {stream_obj.get_uuid()}")
+            stream_dest_dict[name].set(x, k)  # Saving specific data tags 0, 1, 2, ... #record_steps - 1
+
+        # Updating profile
+        if self.is_last_action_step():
+            self.deb("[record] last action step detected, finishing")
+
+            # Dummy get to ensure that the next get will return None (i.e., we only PubSub if somebody restarts this)
+            for stream_obj in stream_dest_dict.values():
+                stream_obj.get(requested_by="send_stream_samples")
+
+            self.add_streams(list(stream_dest_dict.values()), owned=True)
+            self.update_streams_in_profile()
+            await self.subscribe_to_pubsub_owned_streams()
+            await self.send_profile_to_all()
+
+            # New recorded stream
+            self._last_recorded_stream_num += 1
+
+        return True
+
+    async def connect_by_role(self, role: str | list[str], filter_fcn: str | None = None,
+                              time: float = -1., timeout: float = -1.):
+        """Finds and attempts to connect with agents whose profiles match a specific role. It can be optionally
+        filtered by a custom function. It returns True if at least one valid agent is found (async).
+
+        Args:
+            role: The role or list of roles to search for.
+            filter_fcn: The name of an optional filter function.
+            time: The time duration for the action.
+            timeout: The action timeout.
+
+        Returns:
+            True if at least one agent is found and a connection request is made, False otherwise.
+        """
+        self.out(f"Asking to get in touch with all agents whose role is {role}")
+        assert time is not None and timeout is not None, "Missing basic action information"
+
+        if self.get_action_step() == 0:
+            role_list = role if isinstance(role, list) else [role]
+            self._found_agents = set()
+            at_least_one_is_valid = False
+
+            for role in role_list:
+                role = self.ROLE_STR_TO_BITS[role]
+
+                found_addresses1, found_peer_ids1 = self._node_conn.find_addrs_by_role(Agent.ROLE_WORLD_MASTER | role,
+                                                                                       return_peer_ids_too=True)
+                found_addresses2, found_peer_ids2 = self._node_conn.find_addrs_by_role(Agent.ROLE_WORLD_AGENT | role,
+                                                                                       return_peer_ids_too=True)
+                found_addresses = found_addresses1 + found_addresses2
+                found_peer_ids = found_peer_ids1 + found_peer_ids2
+
+                if filter_fcn is not None:
+                    if hasattr(self, filter_fcn):
+                        filter_fcn = getattr(self, filter_fcn)
+                        if callable(filter_fcn):
+                            found_addresses, found_peer_ids = filter_fcn(found_addresses, found_peer_ids)
+                    else:
+                        self.err(f"Filter function not found: {filter_fcn}")
+
+                self.out(f"Found addresses ({len(found_addresses)}) with role: {role}")
+                for f_addr, f_peer_id in zip(found_addresses, found_peer_ids):
+                    if not self._node_conn.is_connected(f_peer_id):
+                        self.out(f"Asking to get in touch with {f_addr}...")
+                        peer_id = await self._node_ask_to_get_in_touch_fcn(addresses=f_addr, public=False)
+                    else:
+                        self.out(f"Not-asking to get in touch with {f_addr}, "
+                                 f"since I am already connected to the corresponding peer...")
+                        peer_id = f_peer_id
+                    if peer_id is not None:
+                        at_least_one_is_valid = True
+                        self._found_agents.add(peer_id)
+                    self.out(f"...returned {peer_id}")
+            return at_least_one_is_valid
+        else:
+            return True
+
+    async def find_agents(self, role: str | list[str], engage: bool = False):
+        """Locally searches through the agent's known peers (world and public agents) to find agents with a specific
+        role. It populates the `_found_agents` set with the peer IDs of matching agents (async).
+
+        Args:
+            role: The role or list of roles to search for.
+            engage: If you want to force the found agents to be the ones that you are engaged with.
+
+        Returns:
+            True if at least one agent is found, False otherwise.
+        """
+        self.out(f"Finding an available agent whose role is {role}")
+        role_list = role if isinstance(role, list) else [role]
+        self._found_agents = set()
+
+        for role_str in role_list:
+            agents = self.all_agents
+            role_int = self.ROLE_STR_TO_BITS[role_str]
+            role_clean = (role_int >> 2) << 2
+            for peer_id, profile in agents.items():
+                _role_int = self.ROLE_STR_TO_BITS[profile.get_dynamic_profile()['connections']['role']]
+                _role_clean = (_role_int >> 2) << 2
+                if _role_clean == role_clean:
+                    self._found_agents.add(peer_id)  # Peer IDs here
+
+        self.deb(f"[find_agents] Found these agents: {self._found_agents}")
+        if engage:
+            self._engaged_agents = copy.deepcopy(self._found_agents)
+        return len(self._found_agents) > 0
+
+    async def next_pref_stream(self):
+        """Moves the internal pointer to the next stream in the list of preferred streams, which is often used for
+        playlist-like operations. It wraps around to the beginning if it reaches the end (async).
+
+        Returns:
+            True if the move is successful, False if the list is empty.
+        """
+        if len(self._preferred_streams) == 0:
+            self.err(f"Cannot move to the next stream because the list of preferred streams is empty")
+            return False
+
+        self._cur_preferred_stream = (self._cur_preferred_stream + 1) % len(self._preferred_streams)
+        suffix = ", warning: restarted" if self._cur_preferred_stream == 0 else ""
+        self.out(f"Moving to the next preferred stream ({self._preferred_streams[self._cur_preferred_stream]}){suffix}")
+        return True
+
+    async def first_pref_stream(self):
+        """Resets the internal pointer to the first stream in the list of preferred streams. This is useful for
+        restarting a playback or processing loop (async).
+
+        Returns:
+            True if the move is successful, False if the list is empty.
+        """
+        if len(self._preferred_streams) == 0:
+            self.err(f"Cannot move to the first stream because the list of preferred streams is empty")
+            return False
+
+        self._cur_preferred_stream = 0
+        self.out(f"Moving to the first preferred stream ({self._preferred_streams[self._cur_preferred_stream]})")
+        return True
+
+    async def check_pref_stream(self, what: str = "last"):
+        """Checks the position of the current preferred stream within the list. It can check if it's the first, last,
+        or if it has completed a full round, among other checks (async).
+
+        Args:
+            what: A string specifying the type of check to perform (e.g., 'first', 'last', 'last_round').
+
+        Returns:
+            True if the condition is met, False otherwise.
+        """
+        valid = ['first', 'last', 'not_first', 'not_last', 'last_round', 'not_last_round', 'last_song', 'not_last_song']
+        assert what in valid, f"The what argument can only be one of {valid}"
+
+        self.out(f"Checking if the current preferred playlist item "
+                 f"(id: {self._cur_preferred_stream}) is the '{what}' one")
+        if what == "first":
+            return self._cur_preferred_stream == 0
+        elif what == "last":
+            return self._cur_preferred_stream == len(self._preferred_streams) - 1
+        elif what == "not_first":
+            return self._cur_preferred_stream != 0
+        elif what == "not_last":
+            return self._cur_preferred_stream != len(self._preferred_streams) - 1
+        elif what == "last_round":
+            return (self._cur_preferred_stream + len(self._preferred_streams) // self._repeat >=
+                    len(self._preferred_streams))
+        elif what == "not_last_round":
+            return (self._cur_preferred_stream + len(self._preferred_streams) // self._repeat <
+                    len(self._preferred_streams))
+        elif what == "last_song":
+            num_streams_in_playlist = len(self._preferred_streams) // self._repeat
+            return (self._cur_preferred_stream + 1) % num_streams_in_playlist == 0
+        elif what == "not_last_song":
+            num_streams_in_playlist = len(self._preferred_streams) // self._repeat
+            return (self._cur_preferred_stream + 1) % num_streams_in_playlist != 0
+
+    async def set_pref_streams(self, net_hashes: list[str], repeat: int = 1):
+        """Fills the agent's list of preferred streams (a playlist). It can repeat the playlist a specified number of
+        times and resolves user-provided stream hashes to their full network hashes (async).
+
+        Args:
+            net_hashes: A list of stream hashes to add to the playlist.
+            repeat: The number of times to repeat the playlist.
+
+        Returns:
+            Always True.
+        """
+        self.out(f"Setting up a list of {len(net_hashes)} preferred streams")
+        self._cur_preferred_stream = 0
+        self._preferred_streams = []
+        self._repeat = repeat
+        for i in range(0, self._repeat):
+            for net_hash in net_hashes:
+
+                # We are tolerating both peer_id:name_or_group and also peer_id::ps:name_or_group
+                components = net_hash.split(":")
+                peer_id = components[0]
+                name_or_group = components[-1]
+                net_hash_to_streams = self.find_streams(peer_id=peer_id, name_or_group=name_or_group)
+                for _net_hash in net_hash_to_streams.keys():
+                    self._preferred_streams.append(_net_hash)
+
+        return True
+
+    async def evaluate(self, stream_hash: str, how: str, steps: int = 100, re_offset: bool = False):
+        """Evaluates the performance of agents that have completed a generation task. It compares the generated data
+        from each agent with a local stream (which can be a ground truth or reference stream) using a specified
+        comparison method (async).
+
+        Args:
+            stream_hash: The hash of the local stream to use for comparison.
+            how: The name of the comparison method to use.
+            steps: The number of steps to perform the evaluation.
+            re_offset: A boolean to indicate whether to re-offset the streams.
+
+        Returns:
+            True if the evaluation is successful, False otherwise.
+        """
+        if not self.buffer_generated_by_others:
+            self.err("Cannot evaluate if not buffering data generated by others")
+            return False
+
+        if stream_hash == "<playlist>":
+            net_hash = self._preferred_streams[self._cur_preferred_stream]
+        else:
+            net_hash = self.user_stream_hash_to_net_hash(stream_hash)
+
+        self._eval_results = {}
+        self.deb(f"[eval] Agents returning streams: {self._agents_who_completed_what_they_were_asked}")
+        for peer_id in self._agents_who_completed_what_they_were_asked:
+            received_net_hash = self.last_buffered_peer_id_to_info[peer_id]["net_hash"]
+            self.out(f"Comparing {net_hash} with {received_net_hash}")
+            eval_result, ret = self.__compare_streams(net_hash_a=net_hash,
+                                                      net_hash_b=received_net_hash,
+                                                      how=how, steps=steps, re_offset=re_offset)
+            self.out(f"Result of the comparison: {eval_result}")
+            if not ret:
+                return False
+            else:
+                peer_id = DataProps.peer_id_from_net_hash(received_net_hash)
+                self._eval_results[peer_id] = eval_result
+
+        return True
+
+    async def compare_eval(self, cmp: str, thres: float, good_if_true: bool = True):
+        """Compares the results of a previous evaluation to a given threshold or finds the best result among all
+        agents. It can check for minimum, maximum, or simple threshold-based comparisons, and it populates a list of
+        'valid' agents that passed the comparison (async).
+
+        Args:
+            cmp: The comparison operator (e.g., '<', '>', 'min').
+            thres: The threshold value for comparison.
+            good_if_true: A boolean to invert the pass/fail logic.
+
+        Returns:
+            True if at least one agent passed the comparison, False otherwise.
+        """
+        assert cmp in ["<", ">", ">=", "<=", "min", "max"], f"Invalid comparison operator: {cmp}"
+        assert thres >= 0. or cmp in ["min", "max"], f"Invalid evaluation threshold: {thres} (it must be in >= 0.)"
+
+        self._valid_cmp_agents = set()
+        msgs = []
+        best_so_far = -1
+
+        min_or_max = None
+        leq_or_geq = None
+        if cmp in ["min", "max"]:
+            min_or_max = "minimum" if cmp == "min" else "maximum"
+            leq_or_geq = "<=" if cmp == "min" else ">="
+
+        for agent_peer_id, eval_result in self._eval_results.items():
+            if cmp not in ["min", "max"]:
+                self.out(f"Checking if result {eval_result} {cmp} {thres}, for agent {agent_peer_id}")
+            else:
+                if thres >= 0:
+                    self.out(f"Checking if result {eval_result} is the {min_or_max} so far, "
+                             f"only if {leq_or_geq} {thres}, for agent {agent_peer_id}")
+                else:
+                    self.out(f"Checking if result {eval_result} is the {min_or_max} so far, for agent {agent_peer_id}")
+
+            if eval_result < 0.:
+                self.err(f"Invalid evaluation result: {eval_result}")
+                return False
+
+            if cmp != "min" and cmp != "max":
+                outcome = False
+                if cmp == "<" and eval_result < thres:
+                    outcome = True
+                elif cmp == "<=" and eval_result <= thres:
+                    outcome = True
+                elif cmp == ">" and eval_result > thres:
+                    outcome = True
+                elif cmp == ">=" and eval_result >= thres:
+                    outcome = True
+
+                if cmp[0] == "<" or cmp[0] == "<=":
+                    alias = 'error level' if good_if_true else 'mark'
+                else:
+                    alias = 'mark' if good_if_true else 'error level'
+
+                if good_if_true:
+                    if outcome:
+                        msgs.append(f"Agent {agent_peer_id} passed with {alias} {eval_result}/{thres}")
+                        self._valid_cmp_agents.add(agent_peer_id)
+                    else:
+                        msgs.append(f"Agent {agent_peer_id} did not pass")
+                else:
+                    if outcome:
+                        msgs.append(f"Agent {agent_peer_id} did not pass")
+                    else:
+                        msgs.append(f"Agent {agent_peer_id} passed with {alias} {eval_result}/{thres}")
+                        self._valid_cmp_agents.add(agent_peer_id)
+
+                if len(msgs) > 1:
+                    msgs[-1] = str(msgs[-1].lower())[0] + msgs[-1][1:]
+            else:
+                if ((cmp == "min" and (thres < 0 or eval_result <= thres) and
+                     (eval_result < best_so_far or best_so_far < 0)) or
+                        (cmp == "max" and (thres < 0 or eval_result >= thres) and
+                         (eval_result > best_so_far or best_so_far < 0))):
+                    best_so_far = eval_result
+                    self._valid_cmp_agents = {agent_peer_id}
+                    msgs = [f"The best agent is {agent_peer_id}"]
+                else:
+                    msgs = [f"No best agent found for the considered threshold ({thres})"]
+
+        if len(self._valid_cmp_agents) == 0:
+
+            # # cheating (hack):
+            # self._valid_cmp_agents.append(agent_peer_id)
+            # self.out(", ".join(msgs))
+            # return True
+            self.err(f"The evaluation was not passed by any agents")
+            return False
+        else:
+            self.out(", ".join(msgs))
+            return True
+    
+    def collect_and_store_own_stats(self):
+        """Collects this agent's own stats and pushes them to the stats recorder."""
+        if self.stats is None:
+            return
+
+        _, own_private_pid = self.get_peer_ids()
+        t = self._node_clock.get_time_ms()
+        try:
+            info = self._node_conn['p2p_world'].get_connected_peers_info()
+            peers_list = [i['id'] for i in info]
+            self.stats.store_stat('connected_peers', peers_list, own_private_pid, t)
+        except Exception:
+            self.stats.store_stat('connected_peers', [], own_private_pid, t)
+
+        try:
+            behav = self.behav
+            self.stats.store_stat('state', behav.get_state_name(), own_private_pid, t)
+            self.stats.store_stat('action', behav.get_action_name(), own_private_pid, t)
+            self.stats.store_stat('last_action', behav.get_last_completed_action_name(), own_private_pid, t)
+        except Exception as e:
+            self.err(f"[Stats] Error storing HSM stats: {e}")
+    
+    async def send_stats_to_world(self):
+        """Sends the agent's currently buffered stats to the world and clears them (async)."""
+        if not self.in_world():
+            self.deb("[send_stats_to_world] Not in a world, skipping stats send.")
+            return
+
+        world_peer_id = self._node_conn.get_world_peer_id()
+        if world_peer_id is None:
+            self.err("[send_stats_to_world] In world, but world_peer_id is None.")
+            return
+
+        self.collect_and_store_own_stats()  # update own stats
+        payload = self.stats.get_payload_for_world()
+        if not payload:
+            self.deb("[send_stats_to_world] No stats to send.")
+            return
+
+        # Send all stats
+        self.out(f"[AGENT] Sending stats update to world {world_peer_id}...")
+        if not (await self._node_conn.send(world_peer_id,
+                                           channel_trail=None,
+                                           content=payload,
+                                           content_type=Msg.STATS_UPDATE)):
+            self.err("Failed to send stats update to world.")
+
+        # Ask the updates to the world (no overwrite required)
+        self.out(f"[AGENT] Requesting stats update from world {world_peer_id}...")
+        if not (await self._node_conn.send(world_peer_id,
+                                           channel_trail=None,
+                                           content={'time_range': self.stats._max_seen_timestamp},
+                                           content_type=Msg.STATS_REQUEST)):
+            self.err("Failed to request stats to world.")
+    
+    def update_stats_view(self, received_view, overwrite: bool = False):
+        """
+        Updates the _world_view attribute of the Stats object.
+        """
+        self.stats.update_view(received_view, overwrite)
+
+    async def suggest_role_to_world(self, agent: str | None, role: str):
+        """Suggests a role change for one or more agents to the world master. It iterates through the involved agents,
+        checks if their current role differs from the suggested one, and sends a role suggestion message to the
+        world master (async).
+
+        Args:
+            agent: The ID of the agent or a wildcard to suggest the role for.
+            role: The new role to suggest (as a string).
+
+        Returns:
+            True if the suggestion was sent successfully, False otherwise.
+        """
+        self.out("Suggesting role to world")
+
+        agents = self.__involved_agents(agent)
+        role_bits = (self.ROLE_STR_TO_BITS[role] >> 2) << 2
+
+        content = []
+
+        for _agent in agents:
+            cur_role_bits = self.ROLE_STR_TO_BITS[self.all_agents[_agent].get_dynamic_profile()['connections']['role']]
+            cur_role_bits = (cur_role_bits >> 2) << 2
+            if cur_role_bits == role_bits:
+                self.out(f"Not suggesting to change the role of {_agent} "
+                         f"since it has already such a role")
+            else:
+                self.out(f"Suggesting to change the role of {_agent} to {self.ROLE_BITS_TO_STR[role_bits]}")
+                content.append({'peer_id': _agent, 'role': role_bits})
+
+        if len(content) > 0:
+            world_peer_id = self._node_conn.get_world_peer_id()
+            if not (await self._node_conn.send(world_peer_id, channel_trail=None,
+                                               content=content,
+                                               content_type=Msg.ROLE_SUGGESTION)):
+                self.err("Failed to send role suggestion to the world")
+                return False
+        return True
+
+    async def suggest_badges_to_world(self, agent: str | None = None,
+                                      score: float = -1.0, badge_type: str = "completed",
+                                      badge_description: str | None = None):
+        """Suggests one or more badges to the world master for specific agents. This is typically used to reward agents
+        for completing tasks, such as for a competition. It sends a message with the badge details, including the score
+        and type, to the world master (async).
+
+        Args:
+            agent: The ID of the agent or a wildcard for which to suggest the badge.
+            score: The score associated with the badge.
+            badge_type: The type of badge (e.g., 'completed').
+            badge_description: An optional description for the badge.
+
+        Returns:
+            True if the badge suggestion was sent successfully, False otherwise.
+        """
+        self.out("Suggesting one or more badges to world")
+
+        if score < 0.:
+            self.err("Invalid score (did you specify the 'score' argument? it must be positive)")
+            return False
+
+        agents = self.__involved_agents(agent)
+        world_peer_id = self._node_conn.get_world_peer_id()
+
+        if badge_type not in Agent.BADGE_TYPES:
+            self.err(f"Unknown badge type: {badge_type}")
+            return False
+
+        list_of_badge_dictionaries = []
+        for peer_id in agents:
+            list_of_badge_dictionaries.append({'peer_id': peer_id,
+                                               'score': score,
+                                               'badge_type': badge_type,
+                                               'badge_description': badge_description,
+                                               'agent_token': self._node_conn.get_last_token(peer_id)})
+
+        if not (await self._node_conn.send(world_peer_id, channel_trail=None,
+                                           content=list_of_badge_dictionaries,
+                                           content_type=Msg.BADGE_SUGGESTIONS)):
+            self.err("Failed to send badge suggestions to the world")
+            return False
+        else:
+            return True
+
+    async def __ask_gen_or_learn(self, for_what: str, agent: str,
+                                 u_hashes: list[str] | None,
+                                 yhat_hashes: list[str] | None,
+                                 samples: int = 100, time: float = -1., timeout: float = -1.,
+                                 ref_uuid: str | None = None):
+        """A private helper method that encapsulates the logic for sending a 'do_gen' or 'do_learn' action request to
+        another agent. It handles the normalization of stream hashes, sets up recipients for direct messages, and adds
+        the target agent to the list of agents asked (async).
+
+        Args:
+            for_what: A string indicating whether to ask for 'gen' or 'learn'.
+            agent: The ID of the agent to send the request to.
+            u_hashes: A list of input stream hashes.
+            yhat_hashes: A list of target stream hashes (for learning).
+            samples: The number of samples.
+            time: The time duration.
+            timeout: The request timeout.
+            ref_uuid: The UUID for the request.
+
+        Returns:
+            True if the request was sent successfully, False otherwise.
+        """
+        if agent not in self.all_agents:
+            self.err(f"Unknown agent: {agent}")
+            return False
+
+        assert for_what in ["gen", "learn"]
+
+        if for_what == "learn":
+            for yhat_hash in yhat_hashes:
+                yhat_stream_dict = self.known_streams[yhat_hash]
+                for yhat_stream in yhat_stream_dict.values():
+                    if isinstance(yhat_stream, BufferedDataStream):
+                        y_text = yhat_stream.to_text_snippet(length=200)
+                        if y_text is not None and len(y_text) > 0:
+                            self.out("Asking to learn: \"" + y_text + "\"")
+
+        # Setting recipient in the case of direct messages
+        # (differently, in case of pubsub, the agent is already sending messages to all)
+        if u_hashes is not None:
+            for u_hash in u_hashes:
+                if not DataProps.is_pubsub_from_net_hash(u_hash):
+                    self.recipients[u_hash] = agent
+        if yhat_hashes is not None:
+            for yhat_hash in yhat_hashes:
+                if not DataProps.is_pubsub_from_net_hash(yhat_hash):
+                    self.recipients[yhat_hash] = agent
+
+        # Triggering
+        if for_what == "gen":
+            if await self.set_next_action(agent, action="do_gen", args={"u_hashes": u_hashes,
+                                                                        "samples": samples, "time": time,
+                                                                        "timeout": timeout},
+                                          ref_uuid=ref_uuid):
+                self._agents_who_were_asked.add(agent)
+                return True
+            else:
+                self.err(f"Unable to ask {agent} to generate")
+                return False
+        elif for_what == "learn":
+            if await self.set_next_action(agent, action="do_learn", args={"u_hashes": u_hashes,
+                                                                          "yhat_hashes": yhat_hashes,
+                                                                          "samples": samples, "time": time,
+                                                                          "timeout": timeout},
+                                          ref_uuid=ref_uuid):
+                self._agents_who_were_asked.add(agent)
+                return True
+            else:
+                self.err(f"Unable to ask {agent} to learn to generate")
+                return False
+
+    def __process_streams(self,
+                          u_hashes: list[str] | None,
+                          yhat_hashes: list[str] | None,
+                          learn: bool = False,
+                          recipient: str | None = None,
+                          ref_uuid: str | None = None):
+        """A private helper method that contains the core logic for processing data streams, either for generation or
+        learning. It reads input streams, passes them to the agent's processor, and handles the output streams.
+        It's designed to be called repeatedly by multistep actions like `do_gen` and `do_learn`.
+
+        Args:
+            u_hashes: A list of input stream hashes.
+            yhat_hashes: A list of target stream hashes (for learning).
+            learn: A boolean to indicate if the task is a learning task.
+            recipient: The ID of the agent to send data back to.
+            ref_uuid: The UUID for the request.
+
+        Returns:
+            True if the stream processing is successful, False otherwise.
+        """
+
+        # Getting current step index
+        k = self.get_action_step()
+
+        # Checking data and creating new buffered streams
+        if k == 0:
+            self.deb("[__process_streams] First action step")
+
+            # Checking data
+            if u_hashes is not None:
+                for u_hash in u_hashes:
+                    if u_hash is not None and u_hash not in self.known_streams:
+                        self.err(f"Unknown stream (u_hash): {u_hash}")
+                        return False
+            if yhat_hashes is not None:
+                for yhat_hash in yhat_hashes:
+                    if yhat_hash is not None and yhat_hash not in self.known_streams:
+                        self.err(f"Unknown stream (yhat_hash): {yhat_hash}")
+                        return False
+
+        if self.is_last_action_step():
+            self.deb("[__process_streams] Last action step detected")
+
+        self.deb(f"[__process_streams] Generating data, step {k}")
+
+        # Generate output
+        outputs, data_tag_from_inputs = (
+            self.generate(input_net_hashes=u_hashes, first=(k == 0), last=self.is_last_action_step(),
+                          ref_uuid=ref_uuid))
+        if outputs is None:
+            return False
+        self.deb(f"[__process_streams] data_tag_from_inputs: {data_tag_from_inputs}")
+        if data_tag_from_inputs is None:
+            data_tag_from_inputs = -1
+            self.deb(f"[__process_streams] data_tag_from_inputs (forced): {data_tag_from_inputs}")
+
+        # Learn
+        if learn:
+            self.deb(f"[__process_streams] learning, step {k}")
+            loss_values, data_tags_from_targets = self.learn_generate(outputs=outputs, targets_net_hashes=yhat_hashes)
+            self.deb(f"[__process_streams] data_tags_from_targets: {data_tags_from_targets}")
+
+            if loss_values is None:
+                return False
+            else:
+                self.out(f"Losses: {loss_values}")
+
+            # Fusing data tags
+            data_tags = [data_tag_from_inputs if _data_tag == -1 else _data_tag for _data_tag in data_tags_from_targets]
+        else:
+            data_tags = [data_tag_from_inputs] * len(outputs)
+        self.deb(f"[__process_streams] data_tags (final): {data_tags}")
+
+        # Set each data sample in "outputs" to the right stream
+        i = 0
+        for net_hash, stream_dict in self.proc_streams.items():
+
+            # Setting the data sample
+            for name, stream_obj in stream_dict.items():
+
+                # Public output streams are only considered if the agent IS NOT acting in a world
+                # private output streams are only considered if the agent IS acting in a world
+                if self.behaving_in_world() != stream_obj.props.is_public():
+
+                    # Guessing recipient of the communication
+                    if i == 0:
+                        self.recipients[net_hash] = recipient \
+                            if not DataProps.is_pubsub_from_net_hash(net_hash) else None
+
+                    self.deb(f"[__process_streams] Setting the {i}-th network output to stream with "
+                             f"net_hash: {net_hash}, name: {name}")
+
+                    # Here we exploit the fact that streams were inserted in order
+                    try:
+                        stream_obj.set(stream_obj.props.check_and_postprocess(outputs[i]), data_tags[i])
+                    except Exception as e:
+                        self.err(f"Error while post-processing the processor output\nException: {e}")
+                        return False
+
+                    if k == 0:
+                        stream_obj.set_uuid(ref_uuid, expected=False)
+                        stream_obj.set_uuid(None, expected=True)
+                    i += 1
+
+        return True
+
+    async def __complete_do(self, do_what: str, peer_id_who_asked: str, all_hashes: list[str] | None,
+                            send_back_confirmation: bool = True):
+        """A private helper method to be called at the end of a `do_gen` or `do_learn` action. It performs cleanup
+        tasks, such as clearing UUIDs on streams, and sends a confirmation message back to the requesting agent (async).
+
+        Args:
+            do_what: A string ('gen' or 'learn') indicating which task was completed.
+            peer_id_who_asked: The ID of the agent who requested the task.
+            all_hashes: A list of all stream hashes involved in the task.
+            send_back_confirmation: A boolean to indicate if a confirmation message should be sent.
+
+        Returns:
+            True if the completion process is successful, False otherwise.
+        """
+        assert do_what in ["gen", "learn"]
+
+        if do_what == "gen":
+            for net_hash, stream_dict in self.proc_streams.items():
+                for stream in stream_dict.values():
+                    if isinstance(stream, BufferedDataStream):
+                        y_text = stream.to_text_snippet(length=200)
+                        if y_text is not None:
+                            self.out("Generated: \"" + y_text + "\"")
+
+        for stream_dict in self.proc_streams.values():
+            for stream_obj in stream_dict.values():
+                if stream_obj.props.is_public() != self.behaving_in_world():
+                    stream_obj.mark_uuid_as_clearable()
+
+        if all_hashes is not None:
+            for net_hash in all_hashes:
+                for stream_obj in self.known_streams[net_hash].values():
+                    stream_obj.set_uuid(None, expected=False)
+                    stream_obj.set_uuid(None, expected=True)
+
+        # Confirming
+        if send_back_confirmation:
+            if await self.set_next_action(peer_id_who_asked, action="done_" + do_what, args={}):
+                return True
+            else:
+                self.err(f"Unable to confirm '{do_what}' to {peer_id_who_asked}")
+                return False
+        else:
+            return True
+
+    def __compare_streams(self, net_hash_a: str, net_hash_b: str,
+                          how: str = "mse", steps: int = 100, re_offset: bool = False):
+        """A private helper method that compares two buffered data streams based on a specified metric (e.g., MSE,
+        max accuracy). It handles stream compatibility checks, data retrieval, and the actual comparison, returning a
+        dissimilarity score.
+
+        Args:
+            net_hash_a: The network hash of the first stream.
+            net_hash_b: The network hash of the second stream.
+            how: The comparison metric ('mse', 'max', 'geqX').
+            steps: The number of samples to compare.
+            re_offset: A boolean to re-align stream tags before comparison.
+
+        Returns:
+            A tuple containing the dissimilarity score and a success flag (e.g., `(0.5, True)`).
+        """
+        if net_hash_a not in self.known_streams:
+            self.err(f"Unknown stream (net_hash_a): {net_hash_a}")
+            return -1., False
+
+        if net_hash_b not in self.known_streams:
+            self.err(f"Unknown stream (net_hash_b): {net_hash_b}")
+            return -1., False
+
+        if steps <= 0:
+            self.err(f"Invalid number of steps: {steps}")
+            return -1., False
+
+        if how not in ["mse", "max"] and not how.startswith("geq"):
+            self.err(f"Data can be compared by MSE, or by comparing the argmax ('max'), or comparing the number "
+                     f"of corresponding bits (obtained by 'geqX', where 'X' is a number). Unknown: {how})")
+            return -1., False
+
+        stream_dict_a = self.known_streams[net_hash_a]
+        stream_dict_b = self.known_streams[net_hash_b]
+
+        if len(stream_dict_a) == 1 and len(stream_dict_b) == 1:
+
+            # If there is only 1 stream is each group, things are easy
+            stream_a = next(iter(stream_dict_a.values()))
+            stream_b = next(iter(stream_dict_b.values()))
+        elif len(stream_dict_a) == 1 and len(stream_dict_b) > 1:
+
+            # If there is only 1 stream is one of the groups, we look for a compatible stream in the other group,
+            # giving priority to streams with labels
+            stream_a = next(iter(stream_dict_a.values()))
+            stream_b = None
+            for stream_obj in stream_dict_b.values():
+                if (stream_a.get_props().has_tensor_labels() and stream_obj.get_props().has_tensor_labels() and
+                        stream_obj.get_props().is_compatible(stream_a.get_props())):
+                    stream_b = stream_obj
+                    break
+            if stream_b is None:
+                for stream_obj in stream_dict_b.values():
+                    if stream_obj.get_props().is_compatible(stream_a.get_props()):
+                        stream_b = stream_obj
+                        break
+        elif len(stream_dict_a) > 1 and len(stream_dict_b) == 1:
+
+            # If there is only 1 stream is one of the groups, we look for a compatible stream in the other group,
+            # giving priority to streams with labels
+            stream_a = None
+            stream_b = next(iter(stream_dict_b.values()))
+            for stream_obj in stream_dict_a.values():
+                if (stream_b.get_props().has_tensor_labels() and stream_obj.get_props().has_tensor_labels() and
+                        stream_obj.get_props().is_compatible(stream_b.get_props())):
+                    stream_a = stream_obj
+                    break
+            if stream_a is None:
+                for stream_obj in stream_dict_a.values():
+                    if stream_obj.get_props().is_compatible(stream_b.get_props()):
+                        stream_a = stream_obj
+                        break
+        else:
+
+            # If both groups have more than a stream, let's give priority to streams with labels to find a match
+            stream_a = None
+            stream_b = None
+            for stream_obj_a in stream_dict_a.values():
+                if not stream_obj_a.get_props().has_tensor_labels():
+                    continue
+                if stream_a is not None and stream_b is not None:
+                    break
+                for stream_obj_b in stream_dict_b.values():
+                    if (stream_obj_b.get_props().has_tensor_labels() and
+                            stream_obj_a.get_props().is_compatible(stream_obj_b.get_props())):
+                        stream_a = stream_obj_a
+                        stream_b = stream_obj_b
+                        break
+            if stream_a is None and stream_b is None:
+                for stream_obj_a in stream_dict_a.values():
+                    if stream_a is not None and stream_b is not None:
+                        break
+                    for stream_obj_b in stream_dict_b.values():
+                        if stream_obj_a.get_props().is_compatible(stream_obj_b.get_props()):
+                            stream_a = stream_obj_a
+                            stream_b = stream_obj_b
+                            break
+
+        if stream_a is None:
+            self.err(f"Cannot find the data stream to consider in the comparison, {net_hash_a}")
+            return -1., False
+        if stream_b is None:
+            self.err(f"Cannot find the data stream to consider in the comparison, {net_hash_b}")
+            return -1., False
+
+        if not isinstance(stream_a, BufferedDataStream):
+            self.err(f"Can only compare buffered streams and {net_hash_a} is not buffered")
+            return -1., False
+
+        if not isinstance(stream_b, BufferedDataStream):
+            self.err(f"Can only compare buffered streams and {net_hash_b} is not buffered")
+            return -1., False
+
+        if steps > len(stream_a) and steps > len(stream_b):
+            self.err(f"Cannot compare streams for {steps} steps, since both of them are shorter "
+                     f"(length of the first stream is {len(stream_a)}, of the second stream is {len(stream_b)})")
+
+        if not stream_a.get_props().is_compatible(stream_b.get_props()):
+            self.err(f"Cannot compare incompatible streams")
+
+        stream_a.restart()
+        stream_b.restart()
+
+        def compare(_a: torch.Tensor | str, _b: torch.Tensor | str, _how: str = "mse") -> float:
+            """Compare two samples of signals or descriptors, returning a dissimilarity score >= 0."""
+
+            assert how in ['mse', 'max', 'same'] or how.startswith("geq"), f"Invalid comparison in terms of {how}"
+
+            if isinstance(_a, torch.Tensor) and isinstance(_b, torch.Tensor):
+                if _a.dtype == torch.long and _b.dtype == torch.long:  # Token IDS
+                    return 1. - float((_a == _b).sum().item()) / a.numel()  # Accuracy
+                elif how == "mse":
+                    ret = torch.nn.functional.mse_loss(_a, _b, reduction='mean')
+                elif how == "max":
+                    ret = 1. - float((torch.argmax(_a) == torch.argmax(_b)).sum().item()) / a.numel()
+                elif how == "same":
+                    ret = 1. - float(torch.eq(_a, _b).sum()) / a.numel()
+                else:
+                    thres = float(how[3:])
+                    ret = 1. - float(torch.sum((_a > thres) == (_b > thres)).item()) / a.numel()
+            else:
+                ret = 1. - float(_a == _b)  # Strings (always handled as 'same')
+            return ret
+
+        # Comparing data (averaging)
+        o = 0.
+        k_b = 0
+        a_tag_offset = 0
+        b_tag_offset = 0
+        a_tag = None
+        a_tag_prev = None
+        for k_a in range(0, steps):
+
+            restart_detected = False
+            if a_tag is not None:
+                a_tag_prev = a_tag
+
+            # Signals or descriptors
+            a, a_tag = stream_a[k_a]
+            b, b_tag = stream_b[k_b]
+
+            # If the streams do not share the same first tag equal to zero, and we asked to re-offset them,
+            # then we force the initial offsets to be zero on both
+            # if not, then re-offset the tags
+            if k_a == 0 and k_b == 0 and re_offset:
+                a_tag_offset = a_tag
+                b_tag_offset = b_tag
+
+            # Offset-based tags
+            a_tag_w_offset = a_tag - a_tag_offset
+            b_tag_w_offset = b_tag - b_tag_offset
+
+            # Checking
+            if a is None:
+                self.err("Cannot compare stream samples if the reference stream yields None")
+                return -1., False
+
+            # Some streams might have been pre-buffered in advance, and have increasing data tags belonging to finite,
+            # fixed set (such as 0, 1, 2, ..., N). when continuously streaming them, we will go from tag N to tag 0 at
+            # a certain point, which is a "restart".
+            # We have to remember that this happened, and we do it for stream "a", our "reference" stream.
+            # Then, below, we will fix tags on stream "b" if needed, considering that such a restart happened.
+            if a_tag_prev is not None and a_tag < a_tag_prev:
+                restart_detected = True
+
+            # Some streams might have been pre-buffered in advance, and have a fixed data tag (usually -1).
+            # Being it negative, it will happen that the data tag will be replaced by a clock cycle, but this function
+            # does not change clock cycles at all, so all samples will have the exact same data tag.
+            # The following code automatically advances the tag by 1 for stream "a", that is expected to be the
+            # reference stream (i.e., the one for which the agent has all samples, with no missing data in between)
+            if a_tag_prev is not None and a_tag <= a_tag_prev:
+                a_tag = a_tag_prev + 1  # Fixed tag detected (patching)
+                a_tag_w_offset = a_tag - a_tag_offset
+
+            # Fixing
+            if b is None:
+                o = o + (1. if how != "mse" else (o / steps) * 1.1)
+                self.deb(f"[__compare_streams] The second stream yields None")
+            else:
+                if b_tag_w_offset == a_tag_w_offset:
+                    o += compare(a, b, how)
+                    k_b += 1
+                    self.deb(f"[__compare_streams] Comparing tags: {a_tag} vs {b_tag} "
+                             f"(with offsets: {a_tag_w_offset} vs {b_tag_w_offset}), samples: {a} vs {b}")
+                elif b_tag_w_offset > a_tag_w_offset:
+                    if not restart_detected:
+                        o = o + (1. if how != "mse" else (o / steps) * 1.1)  # Don't change k_b, some samples missing
+                        self.deb(f"[__compare_streams] (b) Comparing tags: {a_tag} vs {b_tag} -> "
+                                 f"expected one was missing "
+                                 f"(with offsets: {a_tag_w_offset} vs {b_tag_w_offset}) "
+                                 f"samples: {a} vs {b}")
+                    else:
+                        o = o + (1. if how != "mse" else (o / steps) * 1.1)
+                        self.deb(f"[__compare_streams] (c) Comparing tags: {a_tag} vs {b_tag} -> "
+                                 f"expected one was missing "
+                                 f"(with offsets: {a_tag_w_offset} vs {b_tag_w_offset}) "
+                                 f"samples: {a} vs {b}")
+                        k_b += 1  # A restart was detected, it means that "stream_b" is behind, let's move it ahead
+                elif b_tag_w_offset < a_tag_w_offset:
+                    self.deb(f"[__compare_streams] (d) Comparing tags: {a_tag} vs {b_tag} -> too early w.r.t. expected "
+                             f"(with offsets: {a_tag_w_offset} vs {b_tag_w_offset}) "
+                             f"samples: {a} vs {b}")
+                    return -1., False
+
+        self.deb(f"[__compare_streams] Error: {o / steps}")
+
+        # Input("*** press enter to continue ***")
+        return o / steps, True
+
+    def __involved_agents(self, agent: str | None):
+        """A private helper method that resolves an agent ID or a wildcard into a list of specific peer IDs.
+        It can resolve a single agent, a group of agents that passed a previous comparison (`<valid_cmp>`), or all
+        currently engaged agents.
+
+        Args:
+            agent: The agent ID or wildcard string.
+
+        Returns:
+            A list of peer IDs corresponding to the involved agents.
+        """
+        peer_id = agent
+        involved_agents = [peer_id] if peer_id is not None and peer_id != "<valid_cmp>" else (
+            self._valid_cmp_agents) if peer_id is not None and peer_id == "<valid_cmp>" else self._engaged_agents
+        if len(involved_agents) == 0:
+            self.err("Not engaged to any agents or no agent specified")
+        return involved_agents