fix(type-annotations): replace Optional[callable] with Optional[Callable[...]] in 6 files

agent/conversation_loop.py

@@ -25,7 +25,7 @@
 import threading
 import time
 import uuid
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Callable
 
 from agent.codex_responses_adapter import _summarize_user_message_for_log
 from agent.display import KawaiiSpinner
@@ -374,7 +374,7 @@ def run_conversation(
     system_message: str = None,
     conversation_history: List[Dict[str, Any]] = None,
     task_id: str = None,
-    stream_callback: Optional[callable] = None,
+    stream_callback: Optional[Callable[..., None]] = None,
     persist_user_message: Optional[str] = None,
 ) -> Dict[str, Any]:
     """

agent/image_routing.py

@@ -17,12 +17,13 @@
 | ``text``, default ``auto``) and the active model's capability metadata.
 
 In ``auto`` mode:
-  - If the user has explicitly configured ``auxiliary.vision.provider``
+  - If the active model reports ``supports_vision=True`` in its
+    models.dev metadata (or via the ``supports_vision`` config override),
+    we attach natively — the main model can see the pixels directly.
+  - Otherwise, if the user has explicitly configured ``auxiliary.vision.provider``
     (i.e. not ``auto`` and not empty), we assume they want the text pipeline
-    regardless of the main model — they've opted in to a specific vision
-    backend for a reason (cost, quality, local-only, etc.).
-  - Otherwise, if the active model reports ``supports_vision=True`` in its
-    models.dev metadata, we attach natively.
+    as a fallback (they've opted in to a specific vision backend for a reason:
+    cost, quality, local-only, etc.).
   - Otherwise (non-vision model, no explicit override), we fall back to text.
 
 This keeps ``vision_analyze`` surfaced as a tool in every session — skills
@@ -337,12 +338,13 @@ def decide_image_input_mode(
         return "text"
 
     # auto
-    if _explicit_aux_vision_override(cfg):
-        return "text"
-
     supports = _lookup_supports_vision(provider, model, cfg)
     if supports is True:
-        return "native"
+        return "native"  # main model can see images
+
+    if _explicit_aux_vision_override(cfg):
+        return "text"  # main model can't see -> use aux if configured
+
     return "text"
 
 

gateway/stream_consumer.py

@@ -118,7 +118,7 @@ def __init__(
         chat_id: str,
         config: Optional[StreamConsumerConfig] = None,
         metadata: Optional[dict] = None,
-        on_new_message: Optional[callable] = None,
+        on_new_message: Optional[Callable[..., None]] = None,
         initial_reply_to_id: Optional[str] = None,
     ):
         self.adapter = adapter

hermes_cli/dingtalk_auth.py

@@ -17,7 +17,7 @@
 import sys
 import time
 import logging
-from typing import Optional, Tuple
+from typing import Optional, Tuple, Callable
 
 import requests
 
@@ -107,7 +107,7 @@ def wait_for_registration_success(
     device_code: str,
     interval: int = 3,
     expires_in: int = 7200,
-    on_waiting: Optional[callable] = None,
+    on_waiting: Optional[Callable[..., None]] = None,
 ) -> Tuple[str, str]:
     """Block until the registration succeeds or times out.
 

plugins/disk-cleanup/disk_cleanup.py

@@ -26,7 +26,7 @@
 import shutil
 from datetime import datetime, timezone
 from pathlib import Path
-from typing import Any, Dict, List, Optional, Tuple
+from typing import Any, Dict, List, Optional, Tuple, Callable
 
 try:
     from hermes_constants import get_hermes_home
@@ -397,7 +397,7 @@ def quick() -> Dict[str, Any]:
 # ---------------------------------------------------------------------------
 
 def deep(
-    confirm: Optional[callable] = None,
+    confirm: Optional[Callable[[Any], bool]] = None,
 ) -> Dict[str, Any]:
     """Deep cleanup.
 

run_agent.py

@@ -45,7 +45,7 @@
 import time
 import threading
 import uuid
-from typing import List, Dict, Any, Optional
+from typing import List, Dict, Any, Optional, Callable
 # NOTE: `from openai import OpenAI` is deliberately NOT at module top — the
 # SDK pulls ~240 ms of imports. We expose `OpenAI` as a thin proxy object
 # that imports the SDK on first call/isinstance check. This preserves:
@@ -5095,14 +5095,14 @@ def run_conversation(
         system_message: str = None,
         conversation_history: List[Dict[str, Any]] = None,
         task_id: str = None,
-        stream_callback: Optional[callable] = None,
+        stream_callback: Optional[Callable[..., None]] = None,
         persist_user_message: Optional[str] = None,
     ) -> Dict[str, Any]:
         """Forwarder — see ``agent.conversation_loop.run_conversation``."""
         from agent.conversation_loop import run_conversation
         return run_conversation(self, user_message, system_message, conversation_history, task_id, stream_callback, persist_user_message)
 
-    def chat(self, message: str, stream_callback: Optional[callable] = None) -> str:
+    def chat(self, message: str, stream_callback: Optional[Callable[..., None]] = None) -> str:
         """
         Simple chat interface that returns just the final response.
 

tests/agent/test_image_routing.py

@@ -279,15 +279,26 @@ def test_auto_text_for_custom_with_no_override(self):
         with patch("agent.models_dev.get_model_capabilities", return_value=None):
             assert decide_image_input_mode("custom", "unknown", {}) == "text"
 
-    def test_explicit_aux_vision_override_still_wins(self):
-        # If the user has configured a dedicated vision aux backend, respect
-        # it even when supports_vision: true is also set.
+    def test_explicit_aux_vision_override_only_applies_when_main_model_not_vision(self):
+        # Aux override is a fallback: when the main model supports vision,
+        # images go native regardless of aux config. The aux override only
+        # applies when the main model is non-vision.
         cfg = {
             "model": {"supports_vision": True},
             "auxiliary": {"vision": {"provider": "openrouter", "model": "gemini-2.5-pro"}},
         }
         with patch("agent.models_dev.get_model_capabilities", return_value=None):
-            assert decide_image_input_mode("custom", "qwen3.6-35b", cfg) == "text"
+            assert decide_image_input_mode("custom", "qwen3.6-35b", cfg) == "native"
+
+    def test_aux_vision_override_wins_when_main_model_not_vision(self):
+        # When the main model does NOT support vision, the explicit aux
+        # override still forces text routing.
+        cfg = {
+            "model": {"supports_vision": False},
+            "auxiliary": {"vision": {"provider": "openrouter", "model": "gemini-2.5-pro"}},
+        }
+        with patch("agent.models_dev.get_model_capabilities", return_value=None):
+            assert decide_image_input_mode("custom", "text-only-model", cfg) == "text"
 
 
 # ─── build_native_content_parts ──────────────────────────────────────────────

tools/delegate_tool.py

@@ -28,7 +28,7 @@
     ThreadPoolExecutor,
     TimeoutError as FuturesTimeoutError,
 )
-from typing import Any, Dict, List, Optional
+from typing import Any, Dict, List, Optional, Callable
 
 from toolsets import TOOLSETS
 
@@ -725,7 +725,7 @@ def _build_child_progress_callback(
     depth: Optional[int] = None,
     model: Optional[str] = None,
     toolsets: Optional[List[str]] = None,
-) -> Optional[callable]:
+) -> Optional[Callable[..., None]]:
     """Build a callback that relays child agent tool calls to the parent display.
 
     Two display paths: