chore: import upstream snapshot with attribution

examples/showcases/oracle-agent-memory/agent/concierge/memory.py

@@ -0,0 +1,45 @@
+"""Build and cache the OracleAgentMemory client."""
+
+from __future__ import annotations
+
+import functools
+import os
+
+import oracledb
+from oracleagentmemory.core import OracleAgentMemory
+from oracleagentmemory.core.embedders.embedder import Embedder
+from oracleagentmemory.core.llms.llm import Llm
+
+
+def _require(name: str) -> str:
+    value = os.getenv(name)
+    if not value:
+        raise RuntimeError(
+            f"Missing required environment variable {name!r}. "
+            "Copy agent/.env.example to agent/.env and fill it in."
+        )
+    return value
+
+
+@functools.lru_cache(maxsize=1)
+def get_memory() -> OracleAgentMemory:
+    """Return a process-wide singleton memory client (lazily constructed)."""
+    pool = oracledb.create_pool(
+        user=_require("ORACLE_DB_USER"),
+        password=_require("ORACLE_DB_PASSWORD"),
+        dsn=_require("ORACLE_DB_DSN"),  # e.g. "localhost:1521/FREEPDB1"
+        min=1,
+        max=4,
+        increment=1,
+    )
+    embedder = Embedder(model=os.getenv("EMBEDDING_MODEL", "text-embedding-3-small"))
+    llm = Llm(model=os.getenv("MEMORY_LLM_MODEL", "gpt-5.4-mini"))
+    # schema_policy="create_if_necessary" provisions the memory tables on first
+    # run (the cookbook DB user has DB_DEVELOPER_ROLE). Without it, OracleAgentMemory
+    # errors on a fresh database with "Managed DB schema is missing required objects".
+    return OracleAgentMemory(
+        connection=pool,
+        embedder=embedder,
+        llm=llm,
+        schema_policy="create_if_necessary",
+    )