IncludeCPP 4.0.2__py3-none-any.whl → 4.1.0__py3-none-any.whl

includecpp/__init__.py

@@ -2,7 +2,7 @@ from .core.cpp_api import CppApi
 from .core import cssl_bridge as CSSL
 import warnings
 
-__version__ = "4.0.2"
+__version__ = "4.1.0"
 __all__ = ["CppApi", "CSSL"]
 
 # Module-level cache for C++ modules

includecpp/__init__.pyi

@@ -147,7 +147,9 @@ __version__: str
 # Dynamic module access via: from includecpp import <module_name>
 # Auto-generated module declarations
 # These allow: from includecpp import <module_name>
-# (Run 'includecpp rebuild' to generate declarations for your modules)
+combat_sim: Combat_simModuleWrapper
+mcts_engine: Mcts_engineModuleWrapper
+pathfinding: PathfindingModuleWrapper
 
 def __dir__() -> List[str]:
     """List available modules including dynamically loaded C++ modules."""

includecpp/cli/commands.py

@@ -5199,7 +5199,7 @@ def ai_ask(question, module_name, files, all_modules, exclude, think_mode, think
     verbose.phase('init', 'Setting up context')
     verbose.detail('Question', question[:50] + '...' if len(question) > 50 else question)
     verbose.detail('Mode', mode)
-    verbose.detail('Model', ai_mgr.config.get('model', 'gpt-5'))
+    verbose.detail('Model', ai_mgr.config.get('model', 'gpt-4o'))
 
     project_root = Path.cwd()
     plugins_dir = project_root / "plugins"
@@ -5220,7 +5220,7 @@ def ai_ask(question, module_name, files, all_modules, exclude, think_mode, think
         files=len(source_files) + len(plugin_files),
         lines=total_lines,
         tokens=total_lines * 4,  # Approximate
-        model=ai_mgr.config.get('model', 'gpt-5')
+        model=ai_mgr.config.get('model', 'gpt-4o')
     )
 
     if use_websearch:
@@ -6456,7 +6456,7 @@ def _convert_with_ai(content: str, module_name: str, source_file,
         )
 
         if not success:
-            ai_verbose.warning("AI conversion failed, using standard converter")
+            ai_verbose.warning(f"AI conversion failed: {response}")
             ai_verbose.phase('fallback', 'Falling back to standard conversion')
             ai_verbose.end(success=True, message="Fallback successful")
             if to_cpp:

includecpp/core/ai_integration.py

@@ -32,8 +32,12 @@ MODELS = {
     'gpt-3.5-turbo': {'context': 16385, 'endpoint': 'gpt-3.5-turbo'},
     'gpt-4-turbo': {'context': 128000, 'endpoint': 'gpt-4-turbo'},
     'gpt-4o': {'context': 128000, 'endpoint': 'gpt-4o'},
+    'gpt-4o-mini': {'context': 128000, 'endpoint': 'gpt-4o-mini'},
     'gpt-5': {'context': 256000, 'endpoint': 'gpt-5'},
     'gpt-5-nano': {'context': 32000, 'endpoint': 'gpt-5-nano'},
+    'o1': {'context': 200000, 'endpoint': 'o1'},
+    'o1-mini': {'context': 128000, 'endpoint': 'o1-mini'},
+    'o1-preview': {'context': 128000, 'endpoint': 'o1-preview'},
 }
 
 DEFAULT_MODEL = 'gpt-5'
@@ -981,18 +985,43 @@ class AIManager:
             'Content-Type': 'application/json'
         }
         token_limit = min(16000, model_info['context'] // 2)
-        data = {
-            'model': model_info['endpoint'],
-            'messages': [
-                {'role': 'system', 'content': system_prompt},
-                {'role': 'user', 'content': user_prompt}
-            ]
-        }
-        if model.startswith('gpt-5'):
-            data['max_completion_tokens'] = token_limit
+
+        # o1 models use different message format (no system message, combined into user)
+        # gpt-5 models use max_completion_tokens instead of max_tokens
+        is_o1_model = model.startswith('o1')
+        is_gpt5_model = model.startswith('gpt-5')
+
+        if is_o1_model:
+            # o1 models: combine system + user into single user message
+            combined_content = f"{system_prompt}\n\n---\n\n{user_prompt}"
+            data = {
+                'model': model_info['endpoint'],
+                'messages': [
+                    {'role': 'user', 'content': combined_content}
+                ],
+                'max_completion_tokens': token_limit
+            }
+            # o1 models don't support temperature
+        elif is_gpt5_model:
+            # gpt-5 models: use max_completion_tokens, no custom temperature
+            data = {
+                'model': model_info['endpoint'],
+                'messages': [
+                    {'role': 'system', 'content': system_prompt},
+                    {'role': 'user', 'content': user_prompt}
+                ],
+                'max_completion_tokens': token_limit
+            }
         else:
-            data['max_tokens'] = token_limit
-            data['temperature'] = temperature
+            data = {
+                'model': model_info['endpoint'],
+                'messages': [
+                    {'role': 'system', 'content': system_prompt},
+                    {'role': 'user', 'content': user_prompt}
+                ],
+                'max_tokens': token_limit,
+                'temperature': temperature
+            }
         try:
             response = requests.post(OPENAI_API_URL, headers=headers, json=data, timeout=timeout)
             if response.status_code == 200:
@@ -1312,6 +1341,10 @@ class AIManager:
         import platform
         import subprocess
 
+        # Early fail-fast checks
+        if not self.config.get('api_key'):
+            return False, 'No API key configured. Run: includecpp ai setup', []
+
         # Build tools list for prompt
         tools_list = '\n'.join([
             f"- {name}: {info['desc']}\n  Format:\n  {info['format']}"
@@ -1371,9 +1404,9 @@ class AIManager:
 
         prompt = self._build_prompt_with_docs(prompt)
 
-        # Temperature and timeout
+        # Temperature and timeout (reduced for faster failure detection)
         temperature = 0.1 if think_three else (0.2 if think_twice else 0.3)
-        timeout = None if think_three else (420 if think_twice else (300 if think else 180))
+        timeout = 180 if think_three else (120 if think_twice else (90 if think else 60))
 
         # Execute with tool loop
         all_changes = []