chore: import upstream snapshot with attribution

examples/integrations/langgraph-fastapi/agent/main.py

@@ -0,0 +1,55 @@
+import os
+import warnings
+from pathlib import Path
+from dotenv import load_dotenv
+
+# Load .env from the demo project root (one level up from agent/) BEFORE
+# importing src.agent — that import constructs ChatOpenAI at module load,
+# which needs OPENAI_API_KEY in the environment already.
+_demo_root = Path(__file__).parent.parent
+for env_path in (_demo_root / ".env", Path(".env")):
+    if env_path.is_file():
+        load_dotenv(env_path)
+        break
+else:
+    load_dotenv()
+
+from fastapi import FastAPI
+import uvicorn
+from src.agent import graph
+from copilotkit import LangGraphAGUIAgent
+from ag_ui_langgraph import add_langgraph_fastapi_endpoint
+
+app = FastAPI()
+
+
+@app.get("/health")
+async def health():
+    return {"status": "ok"}
+
+
+add_langgraph_fastapi_endpoint(
+    app=app,
+    agent=LangGraphAGUIAgent(
+        name="sample_agent",
+        description="An example agent to use as a starting point for your own agent.",
+        graph=graph,
+    ),
+    path="/",
+)
+
+
+def main():
+    """Run the uvicorn server."""
+    port = int(os.getenv("PORT", "8123"))
+    uvicorn.run(
+        "main:app",
+        host="0.0.0.0",
+        port=port,
+        reload=True,
+    )
+
+
+warnings.filterwarnings("ignore", category=UserWarning, module="pydantic")
+if __name__ == "__main__":
+    main()