autobyteus 1.1.4__py3-none-any.whl → 1.1.5__py3-none-any.whl

examples/workflow/run_debate_workflow.py

@@ -1,212 +0,0 @@
-# file: autobyteus/examples/workflow/run_debate_workflow.py
-"""
-This example script demonstrates a hierarchical workflow.
-A parent workflow (The Debate) manages two sub-workflows (Debating Teams).
-"""
-import asyncio
-import logging
-import argparse
-from pathlib import Path
-import sys
-import os
-
-# --- Boilerplate to make the script runnable from the project root ---
-SCRIPT_DIR = Path(__file__).resolve().parent.parent
-PACKAGE_ROOT = SCRIPT_DIR.parent
-if str(PACKAGE_ROOT) not in sys.path:
-    sys.path.insert(0, str(PACKAGE_ROOT))
-
-# Load environment variables from .env file
-try:
-    from dotenv import load_dotenv
-    load_dotenv(PACKAGE_ROOT / ".env")
-except ImportError:
-    pass
-
-# --- Imports for the Workflow TUI Example ---
-try:
-    from autobyteus.agent.context import AgentConfig
-    from autobyteus.llm.models import LLMModel
-    from autobyteus.llm.llm_factory import default_llm_factory, LLMFactory
-    from autobyteus.workflow.workflow_builder import WorkflowBuilder
-    from autobyteus.cli.workflow_tui.app import WorkflowApp
-    from autobyteus.workflow.context.workflow_config import WorkflowConfig
-except ImportError as e:
-    print(f"Error importing autobyteus components: {e}", file=sys.stderr)
-    sys.exit(1)
-
-# --- Logging Setup ---
-def setup_file_logging() -> Path:
-    log_dir = PACKAGE_ROOT / "logs"
-    log_dir.mkdir(exist_ok=True)
-    log_file_path = log_dir / "debate_workflow_tui_app.log"
-    logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s] %(name)s: %(message)s", filename=log_file_path, filemode="w")
-    logging.getLogger("asyncio").setLevel(logging.WARNING)
-    logging.getLogger("textual").setLevel(logging.WARNING)
-    return log_file_path
-
-def create_debate_workflow(moderator_model: str, affirmative_model: str, negative_model: str, use_xml_tool_format: bool = True):
-    """Creates a hierarchical debate workflow for the TUI demonstration."""
-    # Validate models
-    def _validate_model(model_name: str):
-        try:
-            _ = LLMModel[model_name]
-        except KeyError:
-            logging.critical(f"LLM Model '{model_name}' is not valid. Use --help-models to see available models.")
-            print(f"\nCRITICAL ERROR: LLM Model '{model_name}' is not valid. Use --help-models to see available models.\nCheck log file for details.")
-            sys.exit(1)
-
-    for model in [moderator_model, affirmative_model, negative_model]:
-        _validate_model(model)
-    
-    logging.info(f"Using models -> Moderator: {moderator_model}, Affirmative: {affirmative_model}, Negative: {negative_model}")
-    logging.info(f"Using XML tool format: {use_xml_tool_format}")
-
-    # --- AGENT CONFIGURATIONS ---
-
-    # Parent-Level Agents
-    moderator_config = AgentConfig(
-        name="DebateModerator", role="Coordinator", description="Manages the debate, gives turns, and summarizes.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=moderator_model),
-        system_prompt=(
-            "You are the impartial moderator of a debate between two teams. Your goal is to facilitate a structured, turn-by-turn debate on a user's topic.\n"
-            "Your team consists of Team_Affirmative and Team_Negative. You will delegate tasks to them using their unique names.\n"
-            "Responsibilities: 1. Announce the topic. 2. Ask Team_Affirmative for an opening statement. 3. Ask Team_Negative for a rebuttal. "
-            "4. Facilitate a structured flow of arguments. 5. Conclude the debate.\n"
-            "CRITICAL RULE: You must enforce a strict turn-based system. Only communicate with ONE team at a time using the `SendMessageTo` tool. After sending a message, you must wait for a response before messaging the other team.\n"
-            "Do not debate yourself. Your role is to moderate.\n\n{{tools}}"
-        ),
-        use_xml_tool_format=use_xml_tool_format
-    )
-
-    # Team Affirmative Agents
-    lead_affirmative_config = AgentConfig(
-        name="Lead_Affirmative", role="Coordinator", description="Leads the team arguing FOR the motion.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=affirmative_model),
-        system_prompt=(
-            "You are the lead of the Affirmative team. You receive high-level instructions from the DebateModerator (e.g., 'prepare opening statement').\n"
-            "Your job is to delegate this task to your team member, the Proponent, by giving them a specific instruction.\n\n{{tools}}"
-        ),
-        use_xml_tool_format=use_xml_tool_format
-    )
-    proponent_config = AgentConfig(
-        name="Proponent", role="Debater", description="Argues in favor of the debate topic.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=affirmative_model),
-        system_prompt="You are a Proponent. You will receive instructions from your team lead. Your role is to argue STRONGLY and PERSUASIVELY IN FAVOR of the motion.",
-        use_xml_tool_format=use_xml_tool_format
-    )
-
-    # Team Negative Agents
-    lead_negative_config = AgentConfig(
-        name="Lead_Negative", role="Coordinator", description="Leads the team arguing AGAINST the motion.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=negative_model),
-        system_prompt=(
-            "You are the lead of the Negative team. You receive high-level instructions from the DebateModerator (e.g., 'prepare your rebuttal').\n"
-            "Your job is to delegate this task to your team member, the Opponent, by giving them a specific instruction.\n\n{{tools}}"
-        ),
-        use_xml_tool_format=use_xml_tool_format
-    )
-    opponent_config = AgentConfig(
-        name="Opponent", role="Debater", description="Argues against the debate topic.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=negative_model),
-        system_prompt="You are an Opponent. You will receive instructions from your team lead. Your role is to argue STRONGLY and PERSUASIVELY AGAINST the motion.",
-        use_xml_tool_format=use_xml_tool_format
-    )
-
-    # --- BUILD SUB-WORKFLOWS ---
-    
-    # Build Team Affirmative
-    team_affirmative_workflow: WorkflowConfig = (
-        WorkflowBuilder(name="Team_Affirmative", description="A two-agent team that argues in favor of a proposition.", role="Argues FOR the motion")
-        .set_coordinator(lead_affirmative_config)
-        .add_agent_node(proponent_config)
-        .build()._runtime.context.config # Build to get the config object
-    )
-    
-    # Build Team Negative
-    team_negative_workflow: WorkflowConfig = (
-        WorkflowBuilder(name="Team_Negative", description="A two-agent team that argues against a proposition.", role="Argues AGAINST the motion")
-        .set_coordinator(lead_negative_config)
-        .add_agent_node(opponent_config)
-        .build()._runtime.context.config # Build to get the config object
-    )
-
-    # --- BUILD PARENT WORKFLOW ---
-    
-    debate_workflow = (
-        WorkflowBuilder(name="Grand_Debate", description="A hierarchical workflow for a moderated debate between two teams.")
-        .set_coordinator(moderator_config)
-        .add_workflow_node(team_affirmative_workflow)
-        .add_workflow_node(team_negative_workflow)
-        .build()
-    )
-
-    return debate_workflow
-
-async def main(args: argparse.Namespace, log_file: Path):
-    """Main async function to create the workflow and run the TUI app."""
-    print("Setting up hierarchical debate workflow...")
-    print(f"--> Logs will be written to: {log_file.resolve()}")
-
-    # Resolve model for each role, falling back to the default --llm-model
-    moderator_model = args.moderator_model or args.llm_model
-    affirmative_model = args.affirmative_model or args.llm_model
-    negative_model = args.negative_model or args.llm_model
-    print(f"--> Moderator Model: {moderator_model}")
-    print(f"--> Affirmative Team Model: {affirmative_model}")
-    print(f"--> Negative Team Model: {negative_model}")
-
-    # Determine tool format setting from args
-    use_xml_tool_format = not args.no_xml_tools
-    print(f"--> Using XML Tool Format: {use_xml_tool_format}")
-
-    try:
-        workflow = create_debate_workflow(
-            moderator_model=moderator_model,
-            affirmative_model=affirmative_model,
-            negative_model=negative_model,
-            use_xml_tool_format=use_xml_tool_format,
-        )
-        app = WorkflowApp(workflow=workflow)
-        await app.run_async()
-    except Exception as e:
-        logging.critical(f"Failed to create or run debate workflow TUI: {e}", exc_info=True)
-        print(f"\nCRITICAL ERROR: {e}\nCheck log file for details: {log_file.resolve()}")
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(
-        description="Run a hierarchical 2-team debate workflow with a Textual TUI.",
-        formatter_class=argparse.RawTextHelpFormatter
-    )
-    parser.add_argument("--llm-model", type=str, default="kimi-latest", help="The default LLM model for all agents. Can be overridden by other arguments.")
-    parser.add_argument("--moderator-model", type=str, help="Specific LLM model for the Moderator. Defaults to --llm-model.")
-    parser.add_argument("--affirmative-model", type=str, help="Specific LLM model for the Affirmative Team. Defaults to --llm-model.")
-    parser.add_argument("--negative-model", type=str, help="Specific LLM model for the Negative Team. Defaults to --llm-model.")
-    parser.add_argument("--no-xml-tools", action="store_true", help="Disable XML-based tool formatting. Recommended for models that struggle with XML.")
-    parser.add_argument("--help-models", action="store_true", help="Display available LLM models and exit.")
-    
-    if "--help-models" in sys.argv:
-        try:
-            LLMFactory.ensure_initialized()
-            print("Available LLM Models (you can use either name or value with model arguments):")
-            all_models = sorted(list(LLMModel), key=lambda m: m.name)
-            if not all_models:
-                print("  No models found.")
-            for model in all_models:
-                print(f"  - Name: {model.name:<35} Value: {model.value}")
-        except Exception as e:
-            print(f"Error listing models: {e}")
-        sys.exit(0)
-
-    parsed_args = parser.parse_args()
-
-    log_file_path = setup_file_logging()
-    try:
-        asyncio.run(main(parsed_args, log_file_path))
-    except KeyboardInterrupt:
-        print("\nExiting application.")
-    except Exception as e:
-        # This catches errors during asyncio.run, which might not be logged otherwise
-        logging.critical(f"Top-level application error: {e}", exc_info=True)
-        print(f"\nUNHANDLED ERROR: {e}\nCheck log file for details: {log_file_path.resolve()}")
-

examples/workflow/run_workflow_with_tui.py

@@ -1,153 +0,0 @@
-# file: autobyteus/examples/workflow/run_workflow_with_tui.py
-"""
-This example script demonstrates how to run an AgenticWorkflow with the
-new Textual-based user interface.
-"""
-import asyncio
-import logging
-import argparse
-from pathlib import Path
-import sys
-import os
-
-# --- Boilerplate to make the script runnable from the project root ---
-SCRIPT_DIR = Path(__file__).resolve().parent.parent
-PACKAGE_ROOT = SCRIPT_DIR.parent
-if str(PACKAGE_ROOT) not in sys.path:
-    sys.path.insert(0, str(PACKAGE_ROOT))
-
-# Load environment variables from .env file
-try:
-    from dotenv import load_dotenv
-    load_dotenv(PACKAGE_ROOT / ".env")
-except ImportError:
-    pass
-
-# --- Imports for the Workflow TUI Example ---
-try:
-    from autobyteus.agent.context import AgentConfig
-    from autobyteus.llm.models import LLMModel
-    from autobyteus.llm.llm_factory import default_llm_factory, LLMFactory
-    from autobyteus.workflow.workflow_builder import WorkflowBuilder
-    from autobyteus.cli.workflow_tui.app import WorkflowApp
-except ImportError as e:
-    print(f"Error importing autobyteus components: {e}", file=sys.stderr)
-    sys.exit(1)
-
-# --- Logging Setup ---
-# It's crucial to log to a file so that stdout/stderr are free for Textual.
-def setup_file_logging() -> Path:
-    """
-    Sets up file-based logging and returns the path to the log file.
-    """
-    log_dir = PACKAGE_ROOT / "logs"
-    log_dir.mkdir(exist_ok=True)
-    log_file_path = log_dir / "workflow_tui_app.log"
-    
-    logging.basicConfig(
-        level=logging.INFO,
-        format="%(asctime)s [%(levelname)s] %(name)s: %(message)s",
-        filename=log_file_path,
-        filemode="w",
-    )
-    # Silence the noisy asyncio logger in the file log
-    logging.getLogger("asyncio").setLevel(logging.WARNING)
-    logging.getLogger("textual").setLevel(logging.WARNING)
-    
-    return log_file_path
-
-def create_demo_workflow(model_name: str):
-    """Creates a simple two-agent workflow for the TUI demonstration."""
-    # The factory will handle API key checks based on the selected model's provider.
-
-    # Validate model
-    try:
-        _ = LLMModel[model_name]
-    except KeyError:
-        logging.critical(f"LLM Model '{model_name}' is not valid. Use --help-models to see available models.")
-        print(f"\nCRITICAL ERROR: LLM Model '{model_name}' is not valid. Use --help-models to see available models.\nCheck log file for details.")
-        sys.exit(1)
-
-    # Coordinator Agent Config - Gets its own LLM instance
-    coordinator_config = AgentConfig(
-        name="Coordinator",
-        role="Project Manager",
-        description="Delegates tasks to the team to fulfill the user's request.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=model_name),
-        system_prompt=(
-            "You are a project manager. Your job is to understand the user's request and delegate tasks to your team. "
-            "The workflow will provide you with a team manifest. Use your tools to communicate with your team.\n\n"
-            "Here are your available tools:\n"
-            "{{tools}}"
-        )
-    )
-
-    # Specialist Agent Config (FactChecker) - Gets its own LLM instance
-    fact_checker_config = AgentConfig(
-        name="FactChecker",
-        role="Specialist",
-        description="An agent with a limited, internal knowledge base for answering direct factual questions.",
-        llm_instance=default_llm_factory.create_llm(model_identifier=model_name),
-        system_prompt=(
-            "You are a fact-checking bot. You have the following knowledge:\n"
-            "- The capital of France is Paris.\n"
-            "- The tallest mountain on Earth is Mount Everest.\n"
-            "If asked something you don't know, say 'I do not have information on that topic.'\n\n"
-            "Here is the manifest of tools available to you:\n"
-            "{{tools}}"
-        )
-    )
-
-    # Build the workflow
-    workflow = (
-        WorkflowBuilder(
-            name="TUIDemoWorkflow",
-            description="A simple two-agent workflow for demonstrating the TUI."
-        )
-        .set_coordinator(coordinator_config)
-        .add_agent_node(fact_checker_config, dependencies=[])
-        .build()
-    )
-    return workflow
-
-async def main(args: argparse.Namespace, log_file: Path):
-    """Main async function to create the workflow and run the TUI app."""
-    print("Setting up workflow...")
-    print(f"--> Logs will be written to: {log_file.resolve()}")
-    try:
-        workflow = create_demo_workflow(model_name=args.llm_model)
-        app = WorkflowApp(workflow=workflow)
-        await app.run_async()
-    except Exception as e:
-        logging.critical(f"Failed to create or run workflow TUI: {e}", exc_info=True)
-        print(f"\nCRITICAL ERROR: {e}\nCheck log file for details: {log_file.resolve()}")
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(description="Run an AgenticWorkflow with a Textual TUI.")
-    parser.add_argument("--llm-model", type=str, default="kimi-latest", help="The LLM model to use for the agents.")
-    parser.add_argument("--help-models", action="store_true", help="Display available LLM models and exit.")
-    
-    if "--help-models" in sys.argv:
-        try:
-            LLMFactory.ensure_initialized()
-            print("Available LLM Models (you can use either name or value with --llm-model):")
-            all_models = sorted(list(LLMModel), key=lambda m: m.name)
-            if not all_models:
-                print("  No models found.")
-            for model in all_models:
-                print(f"  - Name: {model.name:<35} Value: {model.value}")
-        except Exception as e:
-            print(f"Error listing models: {e}")
-        sys.exit(0)
-
-    parsed_args = parser.parse_args()
-
-    log_file_path = setup_file_logging()
-    try:
-        asyncio.run(main(parsed_args, log_file_path))
-    except KeyboardInterrupt:
-        print("\nExiting application.")
-    except Exception as e:
-        # This catches errors during asyncio.run, which might not be logged otherwise
-        logging.critical(f"Top-level application error: {e}", exc_info=True)
-        print(f"\nUNHANDLED ERROR: {e}\nCheck log file for details: {log_file_path.resolve()}")