Fix CI: guard starlette imports, asyncio.run(), deprecate datetime.utcnow()

- Guard starlette imports in test_compress_api.py (skip ASGI tests without proxy deps)
- Replace asyncio.get_event_loop().run_until_complete() with asyncio.run() (Python 3.13)
- Replace datetime.utcnow() with datetime.now(timezone.utc).replace(tzinfo=None) everywhere

headroom/cli/memory.py

@@ -6,7 +6,7 @@ import asyncio
 import json
 import os
 import sys
-from datetime import datetime, timedelta
+from datetime import datetime, timedelta, timezone
 from pathlib import Path
 from typing import Any
 
@@ -102,7 +102,7 @@ def _get_stats(store: SQLiteMemoryStore) -> dict[str, Any]:
         stats["by_scope"] = by_scope
 
         # Count by age buckets
-        now = datetime.utcnow()
+        now = datetime.now(timezone.utc).replace(tzinfo=None)
         by_age: dict[str, int] = {}
 
         one_day_ago = (now - timedelta(days=1)).isoformat()
@@ -302,7 +302,7 @@ def list_memories(
 
             if since_duration:
                 duration = parse_duration(since_duration)
-                cutoff = datetime.utcnow() - duration
+                cutoff = datetime.now(timezone.utc).replace(tzinfo=None) - duration
                 filter_kwargs["created_after"] = cutoff
 
             mem_filter = MemoryFilter(**filter_kwargs)
@@ -692,7 +692,7 @@ def prune_memories(
 
         if older_than:
             duration = parse_duration(older_than)
-            cutoff = datetime.utcnow() - duration
+            cutoff = datetime.now(timezone.utc).replace(tzinfo=None) - duration
             filter_kwargs["created_before"] = cutoff
 
         if scope:

headroom/client.py

@@ -3,7 +3,7 @@
 from __future__ import annotations
 
 from collections.abc import Iterator
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import Any
 
 from .cache import (
@@ -388,7 +388,7 @@ class HeadroomClient:
     ) -> Any:
         """Internal implementation of create."""
         request_id = generate_request_id()
-        timestamp = datetime.utcnow()
+        timestamp = datetime.now(timezone.utc).replace(tzinfo=None)
         mode = HeadroomMode(headroom_mode) if headroom_mode else self._default_mode
 
         tokenizer = self._get_tokenizer(model)

headroom/memory/adapters/sqlite.py

@@ -12,7 +12,7 @@ from __future__ import annotations
 import json
 import re
 import sqlite3
-from datetime import datetime
+from datetime import datetime, timezone
 from pathlib import Path
 from typing import Any
 
@@ -606,7 +606,7 @@ class SQLiteMemoryStore:
             ValueError: If the old memory is not found.
         """
         if supersede_time is None:
-            supersede_time = datetime.utcnow()
+            supersede_time = datetime.now(timezone.utc).replace(tzinfo=None)
 
         # Get the old memory
         old_memory = await self.get(old_memory_id)

headroom/memory/backends/mem0.py

@@ -14,7 +14,7 @@ from __future__ import annotations
 import asyncio
 import uuid
 from dataclasses import dataclass
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import Any
 
 from headroom.memory.models import Memory
@@ -217,10 +217,18 @@ class Mem0Backend:
 
         # Parse timestamps
         created_at_str = metadata.get("created_at")
-        created_at = datetime.fromisoformat(created_at_str) if created_at_str else datetime.utcnow()
+        created_at = (
+            datetime.fromisoformat(created_at_str)
+            if created_at_str
+            else datetime.now(timezone.utc).replace(tzinfo=None)
+        )
 
         valid_from_str = metadata.get("valid_from")
-        valid_from = datetime.fromisoformat(valid_from_str) if valid_from_str else datetime.utcnow()
+        valid_from = (
+            datetime.fromisoformat(valid_from_str)
+            if valid_from_str
+            else datetime.now(timezone.utc).replace(tzinfo=None)
+        )
 
         valid_until_str = metadata.get("valid_until")
         valid_until = datetime.fromisoformat(valid_until_str) if valid_until_str else None

headroom/utils.py

@@ -6,7 +6,7 @@ import hashlib
 import json
 import re
 import uuid
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import Any
 
 # Marker format for Headroom modifications
@@ -69,7 +69,7 @@ def compute_prefix_hash(messages: list[dict[str, Any]], prefix_count: int | None
 def format_timestamp(dt: datetime | None = None) -> str:
     """Format datetime as ISO8601 string."""
     if dt is None:
-        dt = datetime.utcnow()
+        dt = datetime.now(timezone.utc).replace(tzinfo=None)
     return dt.isoformat() + "Z"
 
 

tests/test_compress_api.py

@@ -231,9 +231,7 @@ class TestLiteLLMCallback:
             ],
         }
 
-        result = asyncio.get_event_loop().run_until_complete(
-            callback.async_pre_call_hook("key", data, "completion")
-        )
+        result = asyncio.run(callback.async_pre_call_hook("key", data, "completion"))
         assert result is data
 
     def test_callback_ignores_non_completion(self):
@@ -245,7 +243,5 @@ class TestLiteLLMCallback:
         callback = HeadroomCallback()
         data = {"messages": [{"role": "user", "content": "hi"}]}
 
-        result = asyncio.get_event_loop().run_until_complete(
-            callback.async_pre_call_hook("key", data, "embedding")
-        )
+        result = asyncio.run(callback.async_pre_call_hook("key", data, "embedding"))
         assert result is data

tests/test_compression/test_evals.py

@@ -121,7 +121,7 @@ This module provides JWT-based authentication with role-based access control.
 from __future__ import annotations
 
 import logging
-from datetime import datetime, timedelta
+from datetime import datetime, timezone, timedelta
 from typing import Optional, List, Any
 
 import jwt
@@ -181,16 +181,16 @@ def create_access_token(
         >>> print(token)  # eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9...
     """
     if expires_delta:
-        expire = datetime.utcnow() + expires_delta
+        expire = datetime.now(timezone.utc).replace(tzinfo=None) + expires_delta
     else:
-        expire = datetime.utcnow() + timedelta(minutes=TOKEN_EXPIRE_MINUTES)
+        expire = datetime.now(timezone.utc).replace(tzinfo=None) + timedelta(minutes=TOKEN_EXPIRE_MINUTES)
 
     payload = {
         "sub": user_id,
         "exp": expire,
         "roles": roles or [],
         "permissions": permissions or [],
-        "iat": datetime.utcnow(),
+        "iat": datetime.now(timezone.utc).replace(tzinfo=None),
     }
 
     encoded_jwt = jwt.encode(payload, JWT_SECRET, algorithm=JWT_ALGORITHM)
@@ -313,7 +313,7 @@ class RateLimiter:
         Returns:
             True if request is allowed, False if rate limited.
         """
-        now = datetime.utcnow()
+        now = datetime.now(timezone.utc).replace(tzinfo=None)
         cutoff = now - timedelta(seconds=self.window_size)
 
         # Clean old entries

tests/test_compression/test_llm_eval.py

@@ -119,7 +119,7 @@ PRODUCT_CATALOG = json.dumps(
 
 CODE_FILE = '''"""User authentication service with JWT tokens."""
 
-from datetime import datetime, timedelta
+from datetime import datetime, timezone, timedelta
 from typing import Optional
 import jwt
 from pydantic import BaseModel
@@ -155,9 +155,9 @@ def create_access_token(data: dict, expires_delta: Optional[timedelta] = None) -
     """
     to_encode = data.copy()
     if expires_delta:
-        expire = datetime.utcnow() + expires_delta
+        expire = datetime.now(timezone.utc).replace(tzinfo=None) + expires_delta
     else:
-        expire = datetime.utcnow() + timedelta(minutes=ACCESS_TOKEN_EXPIRE_MINUTES)
+        expire = datetime.now(timezone.utc).replace(tzinfo=None) + timedelta(minutes=ACCESS_TOKEN_EXPIRE_MINUTES)
     to_encode.update({"exp": expire})
     encoded_jwt = jwt.encode(to_encode, SECRET_KEY, algorithm=ALGORITHM)
     return encoded_jwt
@@ -215,7 +215,7 @@ class RateLimiter:
 
     def is_allowed(self, client_id: str) -> bool:
         """Check if a request from client_id is allowed."""
-        now = datetime.utcnow()
+        now = datetime.now(timezone.utc).replace(tzinfo=None)
         cutoff = now - timedelta(seconds=self.window_seconds)
 
         if client_id not in self._requests:

tests/test_memory/test_hierarchical.py

@@ -22,7 +22,7 @@ os.environ["TOKENIZERS_PARALLELISM"] = "false"
 
 import asyncio
 import tempfile
-from datetime import datetime, timedelta
+from datetime import datetime, timedelta, timezone
 from pathlib import Path
 
 import numpy as np
@@ -113,7 +113,11 @@ class TestMemoryModel:
         current = Memory(content="test", user_id="alice")
         assert current.is_current is True
 
-        superseded = Memory(content="test", user_id="alice", valid_until=datetime.utcnow())
+        superseded = Memory(
+            content="test",
+            user_id="alice",
+            valid_until=datetime.now(timezone.utc).replace(tzinfo=None),
+        )
         assert superseded.is_current is False
 
     def test_memory_serialization(self, sample_embedding):
@@ -555,7 +559,7 @@ class TestIntegration:
 
         # Supersede
         new_memory = Memory(content="New preference", user_id="alice")
-        supersede_time = datetime.utcnow()
+        supersede_time = datetime.now(timezone.utc).replace(tzinfo=None)
         await store.supersede(original.id, new_memory, supersede_time)
 
         # Query at a point when original was valid (after its valid_from, before supersession)

tests/test_memory_system.py

@@ -17,7 +17,7 @@ import os
 os.environ["TOKENIZERS_PARALLELISM"] = "false"
 
 import tempfile
-from datetime import datetime
+from datetime import datetime, timezone
 from pathlib import Path
 from typing import Any
 
@@ -960,7 +960,7 @@ class MockBackend:
         )
         self._memories[new_memory.id] = new_memory
         old_memory.superseded_by = new_memory.id
-        old_memory.valid_until = datetime.utcnow()
+        old_memory.valid_until = datetime.now(timezone.utc).replace(tzinfo=None)
         return new_memory
 
     async def delete_memory(