aip-agents-binary 0.6.4__py3-none-any.whl

aip_agents/examples/hello_world_model_switch_cli.py

@@ -0,0 +1,210 @@
+"""Hello World Model Switch CLI.
+
+This example demonstrates the enhanced credential system for various LLM providers.
+
+New Auto-Detection Features:
+- Dictionary credentials (e.g., AWS Bedrock)
+- File path credentials (e.g., Google service account JSON)
+- API key strings (e.g., OpenAI, Google API keys)
+
+The system automatically detects credential types - no manual type specification needed!
+
+Setup Environment Variables:
+- BEDROCK_ACCESS_KEY_ID: Your AWS access key for Bedrock
+- BEDROCK_SECRET_ACCESS_KEY: Your AWS secret key for Bedrock
+- GOOGLE_VERTEX_AI_CREDENTIAL_PATH: Path to Google service account JSON file
+- DEEPINFRA_API_KEY: API key for DeepInfra models
+
+Authors:
+    Putu R Wiguna (putu.r.wiguna@gdplabs.id)
+    Christian Trisno Sen Long Chen (christian.t.s.l.chen@gdplabs.id)
+"""
+
+import os
+
+from dotenv import load_dotenv
+from langchain_core.messages import AIMessage, HumanMessage, SystemMessage
+
+from aip_agents.agent import LangChainAgent
+from aip_agents.schema.agent import AgentConfig
+
+load_dotenv()
+
+# You can adjust these model ids to match your backend's supported models
+DEEPINFRA_URL = "https://api.deepinfra.com/v1/openai"
+MODEL_IDS = [
+    "openai/gpt-4.1",
+    "google/gemini-2.5-flash",
+    "bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0",  # Amazon Bedrock example
+    "google/gemini-2.5-pro",  # Google Gemini with service account file
+    "openai-compatible/https://api.deepinfra.com/v1/openai:Qwen/Qwen3-30B-A3B",
+    "openai-compatible/https://api.deepinfra.com/v1/openai:deepseek-ai/DeepSeek-V3",
+    "openai-compatible/https://api.deepinfra.com/v1/openai:deepseek-ai/DeepSeek-R1-0528",
+    # our vllm
+    "openai-compatible/https://ai-agent-vllm.obrol.id/v1/:Qwen/Qwen3-32B-AWQ",  # ensure to turn on vllm-server
+    "openai-compatible/Qwen/Qwen3-30B-A3B",
+    # Azure OpenAI example
+    "azure-openai/https://glair-genai-benchmark.openai.azure.com:glair-benchmark-gpt-4o-mini",
+]
+
+MODEL_CONFIGS = {
+    # Amazon Bedrock - Dictionary credentials (auto-detected)
+    "bedrock/us.anthropic.claude-sonnet-4-20250514-v1:0": AgentConfig(
+        lm_credentials={
+            "access_key_id": os.getenv("BEDROCK_ACCESS_KEY_ID", "your-aws-access-key-id"),
+            "secret_access_key": os.getenv("BEDROCK_SECRET_ACCESS_KEY", "your-aws-secret-access-key"),
+        },
+        lm_hyperparameters={"temperature": 0.7, "maxTokens": 1000},
+    ),
+    # Google Gemini with service account file - File path (auto-detected)
+    "google/gemini-2.5-pro": AgentConfig(
+        lm_credentials=os.getenv("GOOGLE_VERTEX_AI_CREDENTIAL_PATH", "/path/to/service-account.json"),
+        lm_hyperparameters={"temperature": 0.7},
+    ),
+    # Legacy configurations using lm_api_key (still supported)
+    "openai-compatible/https://api.deepinfra.com/v1/openai:Qwen/Qwen3-30B-A3B": AgentConfig(
+        lm_base_url=DEEPINFRA_URL,
+        lm_api_key=os.getenv("DEEPINFRA_API_KEY"),  # Legacy field still works
+    ),
+    "openai-compatible/https://api.deepinfra.com/v1/openai:deepseek-ai/DeepSeek-V3": AgentConfig(
+        lm_base_url=DEEPINFRA_URL,
+        lm_api_key=os.getenv("DEEPINFRA_API_KEY"),
+    ),
+    "openai-compatible/https://api.deepinfra.com/v1/openai:deepseek-ai/DeepSeek-R1-0528": AgentConfig(
+        lm_base_url=DEEPINFRA_URL,
+        lm_api_key=os.getenv("DEEPINFRA_API_KEY"),
+    ),
+    "openai-compatible/https://ai-agent-vllm.obrol.id/v1/:Qwen/Qwen3-32B-AWQ": AgentConfig(
+        lm_base_url="https://ai-agent-vllm.obrol.id/v1/",
+        lm_hyperparameters={"temperature": 1.0},
+    ),
+    "openai-compatible/Qwen/Qwen3-30B-A3B": AgentConfig(
+        lm_base_url=DEEPINFRA_URL,
+        lm_api_key=os.getenv("DEEPINFRA_API_KEY"),
+    ),
+    "openai-compatible/Qwen/Qwen3-32B-AWQ": AgentConfig(
+        lm_base_url="https://ai-agent-vllm.obrol.id/v1/",
+        lm_hyperparameters={"temperature": 1.0},
+    ),
+    "openai/gpt-4.1": AgentConfig(
+        lm_hyperparameters={"temperature": 1.0},
+    ),
+    "azure-openai/https://glair-genai-benchmark.openai.azure.com:glair-benchmark-gpt-4o-mini": AgentConfig(
+        lm_base_url="https://glair-genai-benchmark.openai.azure.com",
+    ),
+}
+
+
+def make_agent(model_id):
+    """Makes an agent with the given model id.
+
+    Args:
+        model_id: The model identifier to use for the agent.
+
+    Returns:
+        LangChainAgent: The configured agent instance.
+    """
+    config = MODEL_CONFIGS.get(model_id)
+    agent = LangChainAgent(
+        name=f"HelloWorldAgent-{model_id}",
+        instruction="You are a helpful assistant.",
+        model=model_id,
+        config=config,
+    )
+    return agent
+
+
+def print_help():
+    """Prints available commands and their descriptions."""
+    help_text = (
+        "\nAvailable commands:\n"
+        "  /help    Show this help message\n"
+        "  /switch  Switch to another model\n"
+        "  /exit    Quit the chat\n"
+        "Type anything else to chat with the assistant.\n"
+    )
+    print(help_text)
+
+
+def handle_switch_model(current_model):
+    """Handles model switching. Returns (new_model, new_agent).
+
+    Args:
+        current_model: The currently active model identifier.
+
+    Returns:
+        tuple: (new_model, new_agent) where new_model is the selected model ID and new_agent is the configured agent instance.
+    """
+    print(f"Current model: {current_model}")
+    print("Available models:")
+    for idx, m in enumerate(MODEL_IDS):
+        print(f"  [{idx + 1}] {m}")
+    try:
+        selection = input("Select model number: ").strip()
+        idx = int(selection) - 1
+        if idx < 0 or idx >= len(MODEL_IDS):
+            print("Invalid selection. No switch performed.")
+            return current_model, make_agent(current_model)
+        new_model = MODEL_IDS[idx]
+    except Exception:
+        print("Invalid input. No switch performed.")
+        return current_model, make_agent(current_model)
+    print(f"Switched to model: {new_model}")
+    return new_model, make_agent(new_model)
+
+
+def main():
+    """Runs the Hello World Model Switch CLI."""
+    print("Welcome to the Hello World Model Switch CLI!")
+    print("🎯 Demonstrating Enhanced Credential Auto-Detection:")
+    print("  • AWS Bedrock with dictionary credentials")
+    print("  • Google Gemini with API key")
+    print("  • Google Gemini with service account file")
+    print("  • Legacy lm_api_key support still works!")
+    print()
+    print(f"Available models: {MODEL_IDS}")
+    current_model = MODEL_IDS[0]
+    agent = make_agent(current_model)
+    print(f"Loaded model: {current_model}")
+    print("Type your message, or type /switch to change model, or /exit to quit or /help to show available commands.")
+
+    # Conversation history as LangChain message objects
+    messages = [SystemMessage(content="You are a helpful assistant.")]
+
+    while True:
+        user_input = input("You: ").strip()
+        if user_input.lower() == "/help":
+            print_help()
+            continue
+        if user_input.lower() == "/exit":
+            print("Goodbye!")
+            break
+        elif user_input.lower() == "/switch":
+            current_model, agent = handle_switch_model(current_model)
+            continue
+        elif user_input == "":
+            continue
+        # Add user message to history
+        messages.append(HumanMessage(content=user_input))
+        # Run the agent synchronously with history
+        try:
+            response = agent.run(query=user_input, messages=messages[:-1])
+            # Extract AI message (should be last in response state)
+            output = response.get("output")
+            ai_message = None
+            # Try to find the returned messages in the response (standard pattern)
+            if "messages" in response and response["messages"]:
+                # Use the last message as the AI reply
+                ai_message = response["messages"][-1]
+            elif output:
+                ai_message = AIMessage(content=output)
+            else:
+                ai_message = AIMessage(content=str(response))
+            messages.append(ai_message)
+            print(f"Agent ({current_model}): {ai_message.content}")
+        except Exception as e:
+            print(f"Error: {e}")
+
+
+if __name__ == "__main__":
+    main()

aip_agents/examples/hello_world_model_switch_cli.pyi

@@ -0,0 +1,30 @@
+from _typeshed import Incomplete
+from aip_agents.agent import LangChainAgent as LangChainAgent
+from aip_agents.schema.agent import AgentConfig as AgentConfig
+
+DEEPINFRA_URL: str
+MODEL_IDS: Incomplete
+MODEL_CONFIGS: Incomplete
+
+def make_agent(model_id):
+    """Makes an agent with the given model id.
+
+    Args:
+        model_id: The model identifier to use for the agent.
+
+    Returns:
+        LangChainAgent: The configured agent instance.
+    """
+def print_help() -> None:
+    """Prints available commands and their descriptions."""
+def handle_switch_model(current_model):
+    """Handles model switching. Returns (new_model, new_agent).
+
+    Args:
+        current_model: The currently active model identifier.
+
+    Returns:
+        tuple: (new_model, new_agent) where new_model is the selected model ID and new_agent is the configured agent instance.
+    """
+def main() -> None:
+    """Runs the Hello World Model Switch CLI."""

aip_agents/examples/hello_world_multi_agent_adk.py

@@ -0,0 +1,75 @@
+"""Multi-agent example using Google ADK with a coordinator agent.
+
+This example demonstrates a coordinator agent that can delegate tasks to specialized agents.
+"""
+
+import asyncio
+
+import nest_asyncio
+
+from aip_agents.agent.google_adk_agent import GoogleADKAgent
+from aip_agents.examples.tools.adk_arithmetic_tools import sum_numbers
+from aip_agents.examples.tools.adk_weather_tool import get_weather
+
+# Apply nest_asyncio to allow nested event loops
+nest_asyncio.apply()
+
+
+async def multi_agent_example():
+    """Demonstrates multi-agent coordination with GoogleADKAgent."""
+    # Create specialized agents
+    weather_agent = GoogleADKAgent(
+        name="WeatherAgent",
+        instruction=(
+            "You are a weather expert. You must use the weather_tool "
+            "to find weather information for a given city. "
+            "Always include the city name in your response."
+        ),
+        model="gemini-2.0-flash",
+        tools=[get_weather],  # Use the get_weather function directly
+        max_iterations=3,
+    )
+
+    math_agent = GoogleADKAgent(
+        name="MathAgent",
+        instruction=(
+            "You are a math expert. You must use the sum_numbers tool to perform addition. "
+            "The tool takes two integer arguments: 'a' and 'b'. "
+            "For example, to add 5 and 7, you would call sum_numbers(a=5, b=7). "
+            "Always state the numbers you're adding in your response."
+        ),
+        model="gemini-2.0-flash",
+        tools=[sum_numbers],
+        max_iterations=3,
+    )
+
+    # Create coordinator agent with access to specialized agents
+    coordinator_agent = GoogleADKAgent(
+        name="CoordinatorAgent",
+        instruction=(
+            "You are a helpful assistant that coordinates between specialized agents.\n"
+            "When asked about weather, delegate to WeatherAgent.\n"
+            "When asked to do math, delegate to MathAgent.\n"
+            "If asked multiple questions, break them down and handle each one separately.\n"
+            "Always be concise and helpful in your responses."
+        ),
+        model="gemini-2.0-flash",
+        agents=[weather_agent, math_agent],
+        max_iterations=3,
+    )
+
+    # Test weather query
+    weather_query = "What is the weather in Tokyo?"
+    print(f"\n--- Running query 1: {weather_query} ---")
+    weather_response = await coordinator_agent.arun(query=weather_query)
+    print(f"Weather Response: {weather_response.get('output')}")
+
+    # Test math query
+    math_query = "What is 5 + 7?"
+    print(f"\n--- Running query 2: {math_query} ---")
+    math_response = await coordinator_agent.arun(query=math_query)
+    print(f"Math Response: {math_response.get('output')}")
+
+
+if __name__ == "__main__":
+    asyncio.run(multi_agent_example())

aip_agents/examples/hello_world_multi_agent_adk.pyi

@@ -0,0 +1,6 @@
+from aip_agents.agent.google_adk_agent import GoogleADKAgent as GoogleADKAgent
+from aip_agents.examples.tools.adk_arithmetic_tools import sum_numbers as sum_numbers
+from aip_agents.examples.tools.adk_weather_tool import get_weather as get_weather
+
+async def multi_agent_example() -> None:
+    """Demonstrates multi-agent coordination with GoogleADKAgent."""

aip_agents/examples/hello_world_multi_agent_langchain.py

@@ -0,0 +1,54 @@
+"""Example demonstrating a multi-agent setup with a Coordinator Agent.
+
+This example showcases:
+1. How to define multiple specialized agents (Weather Agent, Math Agent).
+2. How to set up a Coordinator Agent that can delegate to these specialized agents.
+3. How the Coordinator Agent uses dynamically created tools to call sub-agents.
+4. How the Coordinator Agent can delegate tasks to the appropriate sub-agents.
+
+Authors:
+    Raymond Christopher (raymond.christopher@gdplabs.id)
+    Putu Ravindra Wiguna (putu.r.wiguna@gdplabs.id)
+"""
+
+import asyncio
+
+from langchain_openai import ChatOpenAI
+
+from aip_agents.agent import LangChainAgent
+from aip_agents.examples.tools import weather_tool_langchain as weather_tool
+from aip_agents.examples.tools.langchain_arithmetic_tools import add_numbers
+
+
+async def main():
+    """Main function demonstrating the multi-agent setup with LangChainAgent."""
+    model = ChatOpenAI(model="gpt-4.1", temperature=0)
+
+    weather_agent = LangChainAgent(
+        name="Weather Agent",
+        instruction="You are a weather expert. You must use the get_weather tool to find weather information.",
+        model=model,
+        tools=[weather_tool],
+    )
+
+    math_agent = LangChainAgent(
+        name="Math Agent",
+        instruction="You are a math expert. Use the 'add_numbers' tool to add two numbers.",
+        model=model,
+        tools=[add_numbers],
+    )
+
+    coordinator_agent = LangChainAgent(
+        name="Coordinator Agent",
+        instruction="Delegate each query to suitable agent (Weather Agent or Math Agent) and combine their results.",
+        model=model,
+        agents=[weather_agent, math_agent],
+    )
+
+    query = "What is the weather in Tokyo and what is 5 + 7?"
+    response = await coordinator_agent.arun(query=query)
+    print(f"Response: {response.get('output')}")
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

aip_agents/examples/hello_world_multi_agent_langchain.pyi

@@ -0,0 +1,5 @@
+from aip_agents.agent import LangChainAgent as LangChainAgent
+from aip_agents.examples.tools.langchain_arithmetic_tools import add_numbers as add_numbers
+
+async def main() -> None:
+    """Main function demonstrating the multi-agent setup with LangChainAgent."""

aip_agents/examples/hello_world_multi_agent_langgraph.py

@@ -0,0 +1,66 @@
+"""Example demonstrating a multi-agent setup with a CoordinatorAgent.
+
+This example showcases:
+1. How to define multiple specialized agents (WeatherAgent, MathAgent).
+2. How to set up a CoordinatorAgent that can delegate to these specialized agents.
+3. How the CoordinatorAgent uses dynamically created tools to call sub-agents.
+4. How the CoordinatorAgent can delegate tasks to the appropriate sub-agents.
+
+Authors:
+    Raymond Christopher (raymond.christopher@gdplabs.id)
+    Putu Ravindra Wiguna (putu.r.wiguna@gdplabs.id)
+"""
+
+import asyncio
+
+from langchain_openai import ChatOpenAI
+
+from aip_agents.agent import LangGraphAgent
+from aip_agents.examples.tools import weather_tool_langchain as weather_tool
+from aip_agents.examples.tools.langchain_arithmetic_tools import add_numbers
+
+
+async def main():
+    """Main function demonstrating the multi-agent setup with LangGraphAgent."""
+    llm = ChatOpenAI(model="gpt-4.1", temperature=0)
+
+    # --- Agent Definitions (tools and sub_agents inlined) ---
+    weather_agent = LangGraphAgent(
+        name="WeatherAgent",
+        instruction="You are a weather expert. You must use the get_weather tool to find weather information.",
+        model=llm,
+        tools=[weather_tool],
+    )
+
+    math_agent = LangGraphAgent(
+        name="MathAgent",
+        instruction=(
+            "You are a math expert. You must use the 'add_numbers' tool to perform addition. "
+            "The tool takes two integer arguments: 'a' and 'b'.\n"
+            "For example, to add 5 and 7, you would call add_numbers(a=5, b=7)."
+        ),
+        model=llm,
+        tools=[add_numbers],
+    )
+
+    coordinator_agent = LangGraphAgent(
+        name="CoordinatorAgent",
+        instruction=(
+            "You are a coordinator agent. Your primary role is to delegate tasks to specialized agents. "
+            "Based on the user's query, decide which agent (WeatherAgent or MathAgent) is best suited. "
+            "If a query involves multiple aspects, delegate accordingly. Synthesize their responses."
+        ),
+        model=llm,
+        agents=[weather_agent, math_agent],
+    )
+
+    print("--- Agents Initialized ---")
+
+    query = "What is the weather in Tokyo and what is 5 + 7?"
+    print(f"\n--- Running query: {query} ---")
+    response = await coordinator_agent.arun(query=query)
+    print(f"Response: {response.get('output')}")
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

aip_agents/examples/hello_world_multi_agent_langgraph.pyi

@@ -0,0 +1,5 @@
+from aip_agents.agent import LangGraphAgent as LangGraphAgent
+from aip_agents.examples.tools.langchain_arithmetic_tools import add_numbers as add_numbers
+
+async def main() -> None:
+    """Main function demonstrating the multi-agent setup with LangGraphAgent."""

aip_agents/examples/hello_world_multi_agent_langgraph_lm_invoker.py

@@ -0,0 +1,69 @@
+"""Example demonstrating a multi-agent setup with a CoordinatorAgent.
+
+This example showcases:
+1. How to define multiple specialized agents (WeatherAgent, MathAgent).
+2. How to set up a CoordinatorAgent that can delegate to these specialized agents.
+3. How the CoordinatorAgent uses dynamically created tools to call sub-agents.
+4. How the CoordinatorAgent can delegate tasks to the appropriate sub-agents.
+
+Authors:
+    Raymond Christopher (raymond.christopher@gdplabs.id)
+    Putu Ravindra Wiguna (putu.r.wiguna@gdplabs.id)
+"""
+
+import asyncio
+
+from dotenv import load_dotenv
+
+from aip_agents.agent import LangGraphAgent
+from aip_agents.examples.tools import weather_tool_langchain as weather_tool
+from aip_agents.examples.tools.langchain_arithmetic_tools import add_numbers
+
+load_dotenv(override=True)
+
+
+async def main():
+    """Main function demonstrating the multi-agent setup with Lang."""
+    # llm = ChatOpenAI(model="gpt-4.1", temperature=0)
+    llm = "openai/gpt-4o"
+
+    # --- Agent Definitions (tools and sub_agents inlined) ---
+    weather_agent = LangGraphAgent(
+        name="WeatherAgent",
+        instruction="You are a weather expert. You must use the get_weather tool to find weather information.",
+        model=llm,
+        tools=[weather_tool],
+    )
+
+    math_agent = LangGraphAgent(
+        name="MathAgent",
+        instruction=(
+            "You are a math expert. You must use the 'add_numbers' tool to perform addition. "
+            "The tool takes two integer arguments: 'a' and 'b'.\n"
+            "For example, to add 5 and 7, you would call add_numbers(a=5, b=7)."
+        ),
+        model=llm,
+        tools=[add_numbers],
+    )
+
+    coordinator_agent = LangGraphAgent(
+        name="CoordinatorAgent",
+        instruction=(
+            "You are a coordinator agent. Your primary role is to delegate tasks to specialized agents. "
+            "Based on the user's query, decide which agent (WeatherAgent or MathAgent) is best suited. "
+            "If a query involves multiple aspects, delegate accordingly. Synthesize their responses."
+        ),
+        model=llm,
+        agents=[weather_agent, math_agent],
+    )
+
+    print("--- Agents Initialized ---")
+
+    query = "What is the weather in Tokyo and what is 5 + 7?"
+    print(f"\n--- Running query: {query} ---")
+    response = await coordinator_agent.arun(query=query)
+    print(f"Response: {response.get('output')}")
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

aip_agents/examples/hello_world_multi_agent_langgraph_lm_invoker.pyi

@@ -0,0 +1,5 @@
+from aip_agents.agent import LangGraphAgent as LangGraphAgent
+from aip_agents.examples.tools.langchain_arithmetic_tools import add_numbers as add_numbers
+
+async def main() -> None:
+    """Main function demonstrating the multi-agent setup with Lang."""

aip_agents/examples/hello_world_pii_logger.py

@@ -0,0 +1,21 @@
+"""This example shows how to use the get_logger function to initialize a logger with PII logging enabled.
+
+Authors:
+    Saul Sayers (saul.sayers@gdplabs.id)
+"""
+
+import logging
+
+from aip_agents.utils.logger import get_logger
+
+logger = get_logger(__name__, logging.DEBUG)
+
+# Set up logging format
+sensitive_info = (
+    "contoh nomor ktp 3525011212941001\n"
+    "contoh email john.doe@example.com\n"
+    "contoh nomor telepon +628121729819 dan 0812898029384.\n"
+    "contoh npwp 01.123.456.7-891.234"
+)
+
+logger.info(f"Logging sensitive information for processing: \n{sensitive_info}")

aip_agents/examples/hello_world_pii_logger.pyi

@@ -0,0 +1,5 @@
+from _typeshed import Incomplete
+from aip_agents.utils.logger import get_logger as get_logger
+
+logger: Incomplete
+sensitive_info: str

aip_agents/examples/hello_world_ptc.py

@@ -0,0 +1,49 @@
+"""Minimal PTC hello world example.
+
+Required environment variables:
+- OPENAI_API_KEY
+- E2B_API_KEY
+"""
+
+import asyncio
+
+from aip_agents.agent import LangGraphReactAgent
+from aip_agents.ptc import PromptConfig, PTCSandboxConfig
+
+
+async def main() -> None:
+    """Run a hello-world PTC flow."""
+    instruction = (
+        "You are a helpful assistant with access to execute_ptc_code. "
+        "Use execute_ptc_code to run Python and print output. "
+        "The tool returns JSON with ok/stdout/stderr/exit_code."
+    )
+
+    agent = LangGraphReactAgent(
+        name="ptc_hello_world",
+        instruction=instruction,
+        model="openai/gpt-5.2",
+        ptc_config=PTCSandboxConfig(enabled=True, sandbox_timeout=180.0, prompt=PromptConfig(mode="index")),
+    )
+    agent.add_mcp_server(
+        {
+            "deepwiki": {
+                "transport": "streamable-http",
+                "url": "https://mcp.deepwiki.com/mcp",
+                "headers": {},
+                "timeout": 60.0,
+            }
+        }
+    )
+
+    try:
+        response = await agent.arun(
+            query="Use execute_ptc_code to print 'Hello, world!' and count the number of words in the output of deepwiki.read_wiki_structure('anthropics/claude-code')."
+        )
+        print("execute_ptc_code output:", response["output"])
+    finally:
+        await agent.cleanup()
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

aip_agents/examples/hello_world_ptc.pyi

@@ -0,0 +1,5 @@
+from aip_agents.agent import LangGraphReactAgent as LangGraphReactAgent
+from aip_agents.ptc import PTCSandboxConfig as PTCSandboxConfig, PromptConfig as PromptConfig
+
+async def main() -> None:
+    """Run a hello-world PTC flow."""