[AI-262] Update and refactor example code

agents-core/vision_agents/core/agents/agents.py

@@ -5,12 +5,11 @@
 import time
 import uuid
 from dataclasses import asdict
-from typing import TYPE_CHECKING, Any, Dict, List, Optional, TypeGuard, Coroutine
+from typing import TYPE_CHECKING, Any, Dict, List, Optional, TypeGuard
 from uuid import uuid4
 
 import getstream.models
 from aiortc import VideoStreamTrack
-from getstream.video.async_call import Call
 from getstream.video.rtc import Call
 
 from getstream.video.rtc.pb.stream.video.sfu.models.models_pb2 import TrackType
@@ -697,7 +696,7 @@ async def create_user(self) -> None:
     async def create_call(self, call_type: str, call_id: str) -> Call:
         """Shortcut for creating a call/room etc."""
         call = self.edge.client.video.call(call_type, call_id)
-        response = await call.get_or_create(data={"created_by_id": self.agent_user.id})
+        await call.get_or_create(data={"created_by_id": self.agent_user.id})
 
         return call
 

examples/01_simple_agent_example/simple_agent_example.py

@@ -1,15 +1,18 @@
-import asyncio
-from uuid import uuid4
+import logging
+
 from dotenv import load_dotenv
 
-from vision_agents.core import User, Agent
+from vision_agents.core import User, Agent, cli
+from vision_agents.core.agents import AgentLauncher
 from vision_agents.plugins import deepgram, getstream, gemini, vogent, elevenlabs
 # from vision_agents.core.profiling import Profiler
 
+logger = logging.getLogger(__name__)
+
 load_dotenv()
 
 
-async def start_agent() -> None:
+async def create_agent(**kwargs) -> Agent:
     llm = gemini.LLM("gemini-2.0-flash")
     # create an agent to run with Stream's edge, openAI llm
     agent = Agent(
@@ -30,9 +33,14 @@ async def start_agent() -> None:
         # realtime version (vad, tts and stt not needed)
         # llm=openai.Realtime()
     )
+    return agent
+
 
+async def join_call(agent: Agent, call_type: str, call_id: str, **kwargs) -> None:
+    # ensure the agent user is created
+    await agent.create_user()
     # Create a call
-    call = agent.edge.client.video.call("default", str(uuid4()))
+    call = await agent.create_call(call_type, call_id)
 
     # Have the agent join the call/room
     with await agent.join(call):
@@ -89,4 +97,4 @@ def _flush_and_shutdown():
 
 if __name__ == "__main__":
     # setup_telemetry()
-    asyncio.run(start_agent())
+    cli(AgentLauncher(create_agent=create_agent, join_call=join_call))

examples/02_golf_coach_example/golf_coach_example.py

@@ -1,15 +1,17 @@
-import asyncio
-from uuid import uuid4
+import logging
 
 from dotenv import load_dotenv
 
-from vision_agents.core import User, Agent
+from vision_agents.core import User, Agent, cli
+from vision_agents.core.agents import AgentLauncher
 from vision_agents.plugins import getstream, ultralytics, gemini
 
+logger = logging.getLogger(__name__)
+
 load_dotenv()
 
 
-async def start_agent() -> None:
+async def create_agent(**kwargs) -> Agent:
     agent = Agent(
         edge=getstream.Edge(),  # use stream for edge video transport
         agent_user=User(name="AI golf coach"),
@@ -20,9 +22,14 @@ async def start_agent() -> None:
             ultralytics.YOLOPoseProcessor(model_path="yolo11n-pose.pt")
         ],  # realtime pose detection with yolo
     )
+    return agent
+
 
-    # create a call, some other video networks call this a room
-    call = agent.edge.client.video.call("default", str(uuid4()))
+async def join_call(agent: Agent, call_type: str, call_id: str, **kwargs) -> None:
+    # ensure the agent user is created
+    await agent.create_user()
+    # Create a call
+    call = await agent.create_call(call_type, call_id)
 
     # join the call and open a demo env
     with await agent.join(call):
@@ -37,4 +44,4 @@ async def start_agent() -> None:
 
 
 if __name__ == "__main__":
-    asyncio.run(start_agent())
+    cli(AgentLauncher(create_agent=create_agent, join_call=join_call))

examples/other_examples/09_github_mcp_demo/gemini_realtime_github_mcp_demo.py

@@ -5,13 +5,13 @@
 using voice commands through the Gemini Live API.
 """
 
-import asyncio
 import logging
 import os
-from uuid import uuid4
+
 from dotenv import load_dotenv
 
-from vision_agents.core.agents import Agent
+from vision_agents.core.agents import Agent, AgentLauncher
+from vision_agents.core import cli
 from vision_agents.core.mcp import MCPServerRemote
 from vision_agents.plugins.gemini.gemini_realtime import Realtime
 from vision_agents.plugins import getstream
@@ -26,22 +26,22 @@
 logger = logging.getLogger(__name__)
 
 
-async def start_agent():
+async def create_agent(**kwargs) -> Agent:
     """Demonstrate Gemini Realtime with GitHub MCP server integration."""
 
     # Get GitHub PAT from environment
     github_pat = os.getenv("GITHUB_PAT")
     if not github_pat:
         logger.error("GITHUB_PAT environment variable not found!")
         logger.error("Please set GITHUB_PAT in your .env file or environment")
-        return
+        raise ValueError("GITHUB_PAT environment variable not found")
 
     # Get Google API key from environment
     google_api_key = os.getenv("GOOGLE_API_KEY")
     if not google_api_key:
         logger.error("GOOGLE_API_KEY environment variable not found!")
         logger.error("Please set GOOGLE_API_KEY in your .env file or environment")
-        return
+        raise ValueError("GOOGLE_API_KEY environment variable not found")
 
     # Create GitHub MCP server
     github_server = MCPServerRemote(
@@ -69,6 +69,10 @@ async def start_agent():
     logger.info("Agent created with Gemini Realtime and GitHub MCP server")
     logger.info(f"GitHub server: {github_server}")
 
+    return agent
+
+
+async def join_call(agent: Agent, call_type: str, call_id: str, **kwargs) -> None:
     try:
         # Set up event handler for when participants join
         @agent.subscribe
@@ -85,8 +89,10 @@ async def on_participant_joined(event: CallSessionParticipantJoinedEvent):
                 f"Hello {event.participant.user.name}! I'm your GitHub AI assistant powered by Gemini Live. I have access to {len(mcp_functions)} GitHub tools and can help you with repositories, issues, pull requests, and more through voice commands!"
             )
 
+        # ensure the agent user is created
+        await agent.create_user()
         # Create a call
-        call = agent.edge.client.video.call("default", str(uuid4()))
+        call = await agent.create_call(call_type, call_id)
 
         # Have the agent join the call/room
         logger.info("🎤 Agent joining call...")
@@ -123,4 +129,4 @@ async def on_participant_joined(event: CallSessionParticipantJoinedEvent):
 
 
 if __name__ == "__main__":
-    asyncio.run(start_agent())
+    cli(AgentLauncher(create_agent=create_agent, join_call=join_call))

examples/other_examples/09_github_mcp_demo/github_mcp_demo.py

@@ -5,13 +5,13 @@
 by the LLM without any manual registration required.
 """
 
-import asyncio
 import logging
 import os
-from uuid import uuid4
+
 from dotenv import load_dotenv
 
-from vision_agents.core.agents import Agent
+from vision_agents.core.agents import Agent, AgentLauncher
+from vision_agents.core import cli
 from vision_agents.core.mcp import MCPServerRemote
 from vision_agents.plugins.openai.openai_llm import OpenAILLM
 from vision_agents.plugins import elevenlabs, deepgram, silero, getstream
@@ -26,15 +26,15 @@
 logger = logging.getLogger(__name__)
 
 
-async def start_agent():
+async def create_agent(**kwargs) -> Agent:
     """Demonstrate GitHub MCP server integration."""
 
     # Get GitHub PAT from environment
     github_pat = os.getenv("GITHUB_PAT")
     if not github_pat:
         logger.error("GITHUB_PAT environment variable not found!")
         logger.error("Please set GITHUB_PAT in your .env file or environment")
-        return
+        raise ValueError("GITHUB_PAT environment variable not found")
 
     # Create GitHub MCP server
     github_server = MCPServerRemote(
@@ -49,7 +49,7 @@ async def start_agent():
     if not openai_api_key:
         logger.error("OPENAI_API_KEY environment variable not found!")
         logger.error("Please set OPENAI_API_KEY in your .env file or environment")
-        return
+        raise ValueError("OPENAI_API_KEY environment variable not found")
 
     # Create OpenAI LLM
     llm = OpenAILLM(model="gpt-4o", api_key=openai_api_key)
@@ -74,6 +74,10 @@ async def start_agent():
     logger.info("Agent created with GitHub MCP server")
     logger.info(f"GitHub server: {github_server}")
 
+    return agent
+
+
+async def join_call(agent: Agent, call_type: str, call_id: str, **kwargs) -> None:
     try:
         # Connect to GitHub MCP server with timeout
         logger.info("Connecting to GitHub MCP server...")
@@ -95,8 +99,10 @@ async def on_participant_joined(event: CallSessionParticipantJoinedEvent):
                 f"Hello {event.participant.user.name}! I'm your GitHub AI assistant with access to {len(mcp_functions)} GitHub tools. I can help you with repositories, issues, pull requests, and more!"
             )
 
+        # ensure the agent user is created
+        await agent.create_user()
         # Create a call
-        call = agent.edge.client.video.call("default", str(uuid4()))
+        call = await agent.create_call(call_type, call_id)
 
         # Have the agent join the call/room
         logger.info("🎤 Agent joining call...")
@@ -126,4 +132,4 @@ async def on_participant_joined(event: CallSessionParticipantJoinedEvent):
 
 
 if __name__ == "__main__":
-    asyncio.run(start_agent())
+    cli(AgentLauncher(create_agent=create_agent, join_call=join_call))

examples/other_examples/09_github_mcp_demo/openai_realtime_github_mcp_demo.py

@@ -5,13 +5,13 @@
 using voice commands through the OpenAI Realtime API.
 """
 
-import asyncio
 import logging
 import os
-from uuid import uuid4
+
 from dotenv import load_dotenv
 
-from vision_agents.core.agents import Agent
+from vision_agents.core.agents import Agent, AgentLauncher
+from vision_agents.core import cli
 from vision_agents.core.mcp import MCPServerRemote
 from vision_agents.plugins.openai.openai_realtime import Realtime
 from vision_agents.plugins import getstream
@@ -26,22 +26,22 @@
 logger = logging.getLogger(__name__)
 
 
-async def start_agent():
+async def create_agent(**kwargs) -> Agent:
     """Demonstrate OpenAI Realtime with GitHub MCP server integration."""
 
     # Get GitHub PAT from environment
     github_pat = os.getenv("GITHUB_PAT")
     if not github_pat:
         logger.error("GITHUB_PAT environment variable not found!")
         logger.error("Please set GITHUB_PAT in your .env file or environment")
-        return
+        raise ValueError("GITHUB_PAT environment variable not found")
 
     # Check OpenAI API key from environment
     openai_api_key = os.getenv("OPENAI_API_KEY")
     if not openai_api_key:
         logger.error("OPENAI_API_KEY environment variable not found!")
         logger.error("Please set OPENAI_API_KEY in your .env file or environment")
-        return
+        raise ValueError("OPENAI_API_KEY environment variable not found")
 
     # Create GitHub MCP server
     github_server = MCPServerRemote(
@@ -71,6 +71,10 @@ async def start_agent():
     logger.info("Agent created with OpenAI Realtime and GitHub MCP server")
     logger.info(f"GitHub server: {github_server}")
 
+    return agent
+
+
+async def join_call(agent: Agent, call_type: str, call_id: str, **kwargs) -> None:
     try:
         # Set up event handler for when participants join
         @agent.subscribe
@@ -87,8 +91,10 @@ async def on_participant_joined(event: CallSessionParticipantJoinedEvent):
                 f"Hello {event.participant.user.name}! I'm your GitHub AI assistant powered by OpenAI Realtime. I have access to {len(mcp_functions)} GitHub tools and can help you with repositories, issues, pull requests, and more through voice commands!"
             )
 
+        # ensure the agent user is created
+        await agent.create_user()
         # Create a call
-        call = agent.edge.client.video.call("default", str(uuid4()))
+        call = await agent.create_call(call_type, call_id)
 
         # Have the agent join the call/room
         logger.info("🎤 Agent joining call...")
@@ -125,4 +131,4 @@ async def on_participant_joined(event: CallSessionParticipantJoinedEvent):
 
 
 if __name__ == "__main__":
-    asyncio.run(start_agent())
+    cli(AgentLauncher(create_agent=create_agent, join_call=join_call))

examples/other_examples/gemini_live_realtime/gemini_live_example.py

@@ -1,12 +1,10 @@
-import asyncio
 import logging
-from uuid import uuid4
 
 from dotenv import load_dotenv
-from getstream import AsyncStream
 
 from vision_agents.core.edge.types import User
-from vision_agents.core.agents import Agent
+from vision_agents.core.agents import Agent, AgentLauncher
+from vision_agents.core import cli
 from vision_agents.plugins import gemini, getstream
 
 load_dotenv()
@@ -18,9 +16,7 @@
 logger = logging.getLogger(__name__)
 
 
-async def start_agent() -> None:
-    client = AsyncStream()
-
+async def create_agent(**kwargs) -> Agent:
     agent = Agent(
         edge=getstream.Edge(),
         agent_user=User(
@@ -30,17 +26,20 @@ async def start_agent() -> None:
         llm=gemini.Realtime(),
         processors=[],  # processors can fetch extra data, check images/audio data or transform video
     )
-
-    call = client.video.call("default", str(uuid4()))
+    return agent
 
 
+async def join_call(agent: Agent, call_type: str, call_id: str, **kwargs) -> None:
+    # ensure the agent user is created
+    await agent.create_user()
+    # Create a call
+    call = await agent.create_call(call_type, call_id)
 
     with await agent.join(call):
-        await asyncio.sleep(5)
         await agent.edge.open_demo(call)
         await agent.llm.simple_response(text="Describe what you see and say hi")
         await agent.finish()  # run till the call ends
 
 
 if __name__ == "__main__":
-    asyncio.run(start_agent())
+    cli(AgentLauncher(create_agent=create_agent, join_call=join_call))