deepcode-hku 1.0.7__tar.gz → 1.0.8__tar.gz

{deepcode_hku-1.0.7/deepcode_hku.egg-info → deepcode_hku-1.0.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: deepcode-hku
-Version: 1.0.7
+Version: 1.0.8
 Summary: AI Research Engine - Transform research papers into working code automatically
 Home-page: https://github.com/HKUDS/DeepCode
 Author: DeepCodeTeam
@@ -24,6 +24,7 @@ Requires-Dist: aiohttp>=3.8.0
 Requires-Dist: anthropic
 Requires-Dist: asyncio-mqtt
 Requires-Dist: docling
+Requires-Dist: google-genai
 Requires-Dist: mcp-agent
 Requires-Dist: mcp-server-git
 Requires-Dist: nest_asyncio
@@ -587,6 +588,14 @@ curl -O https://raw.githubusercontent.com/HKUDS/DeepCode/main/mcp_agent.secrets.
 # Edit mcp_agent.secrets.yaml with your API keys and base_url:
 # - openai: api_key, base_url (for OpenAI/custom endpoints)
 # - anthropic: api_key (for Claude models)
+# - google: api_key (for Gemini models)
+
+# 🤖 Select your preferred LLM provider (optional)
+# Edit mcp_agent.config.yaml to choose your LLM (line ~106):
+# - llm_provider: "google"    # Use Google Gemini models
+# - llm_provider: "anthropic" # Use Anthropic Claude models
+# - llm_provider: "openai"    # Use OpenAI/compatible models
+# Note: If not set or unavailable, will automatically fallback to first available provider
 
 # 🔑 Configure search API keys for web search (optional)
 # Edit mcp_agent.config.yaml to set your API keys:
@@ -623,6 +632,14 @@ uv pip install -r requirements.txt
 # Edit mcp_agent.secrets.yaml with your API keys and base_url:
 # - openai: api_key, base_url (for OpenAI/custom endpoints)
 # - anthropic: api_key (for Claude models)
+# - google: api_key (for Gemini models)
+
+# 🤖 Select your preferred LLM provider (optional)
+# Edit mcp_agent.config.yaml to choose your LLM (line ~106):
+# - llm_provider: "google"    # Use Google Gemini models
+# - llm_provider: "anthropic" # Use Anthropic Claude models
+# - llm_provider: "openai"    # Use OpenAI/compatible models
+# Note: If not set or unavailable, will automatically fallback to first available provider
 
 # 🔑 Configure search API keys for web search (optional)
 # Edit mcp_agent.config.yaml to set your API keys:
@@ -649,6 +666,14 @@ pip install -r requirements.txt
 # Edit mcp_agent.secrets.yaml with your API keys and base_url:
 # - openai: api_key, base_url (for OpenAI/custom endpoints)
 # - anthropic: api_key (for Claude models)
+# - google: api_key (for Gemini models)
+
+# 🤖 Select your preferred LLM provider (optional)
+# Edit mcp_agent.config.yaml to choose your LLM (line ~106):
+# - llm_provider: "google"    # Use Google Gemini models
+# - llm_provider: "anthropic" # Use Anthropic Claude models
+# - llm_provider: "openai"    # Use OpenAI/compatible models
+# Note: If not set or unavailable, will automatically fallback to first available provider
 
 # 🔑 Configure search API keys for web search (optional)
 # Edit mcp_agent.config.yaml to set your API keys:

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8}/README.md

@@ -542,6 +542,14 @@ curl -O https://raw.githubusercontent.com/HKUDS/DeepCode/main/mcp_agent.secrets.
 # Edit mcp_agent.secrets.yaml with your API keys and base_url:
 # - openai: api_key, base_url (for OpenAI/custom endpoints)
 # - anthropic: api_key (for Claude models)
+# - google: api_key (for Gemini models)
+
+# 🤖 Select your preferred LLM provider (optional)
+# Edit mcp_agent.config.yaml to choose your LLM (line ~106):
+# - llm_provider: "google"    # Use Google Gemini models
+# - llm_provider: "anthropic" # Use Anthropic Claude models
+# - llm_provider: "openai"    # Use OpenAI/compatible models
+# Note: If not set or unavailable, will automatically fallback to first available provider
 
 # 🔑 Configure search API keys for web search (optional)
 # Edit mcp_agent.config.yaml to set your API keys:
@@ -578,6 +586,14 @@ uv pip install -r requirements.txt
 # Edit mcp_agent.secrets.yaml with your API keys and base_url:
 # - openai: api_key, base_url (for OpenAI/custom endpoints)
 # - anthropic: api_key (for Claude models)
+# - google: api_key (for Gemini models)
+
+# 🤖 Select your preferred LLM provider (optional)
+# Edit mcp_agent.config.yaml to choose your LLM (line ~106):
+# - llm_provider: "google"    # Use Google Gemini models
+# - llm_provider: "anthropic" # Use Anthropic Claude models
+# - llm_provider: "openai"    # Use OpenAI/compatible models
+# Note: If not set or unavailable, will automatically fallback to first available provider
 
 # 🔑 Configure search API keys for web search (optional)
 # Edit mcp_agent.config.yaml to set your API keys:
@@ -604,6 +620,14 @@ pip install -r requirements.txt
 # Edit mcp_agent.secrets.yaml with your API keys and base_url:
 # - openai: api_key, base_url (for OpenAI/custom endpoints)
 # - anthropic: api_key (for Claude models)
+# - google: api_key (for Gemini models)
+
+# 🤖 Select your preferred LLM provider (optional)
+# Edit mcp_agent.config.yaml to choose your LLM (line ~106):
+# - llm_provider: "google"    # Use Google Gemini models
+# - llm_provider: "anthropic" # Use Anthropic Claude models
+# - llm_provider: "openai"    # Use OpenAI/compatible models
+# Note: If not set or unavailable, will automatically fallback to first available provider
 
 # 🔑 Configure search API keys for web search (optional)
 # Edit mcp_agent.config.yaml to set your API keys:

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8}/__init__.py

@@ -5,7 +5,7 @@ DeepCode - AI Research Engine
 ⚡ Transform research papers into working code automatically
 """
 
-__version__ = "1.0.7"
+__version__ = "1.0.8"
 __author__ = "DeepCode Team"
 __url__ = "https://github.com/HKUDS/DeepCode"
 

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8/deepcode_hku.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: deepcode-hku
-Version: 1.0.7
+Version: 1.0.8
 Summary: AI Research Engine - Transform research papers into working code automatically
 Home-page: https://github.com/HKUDS/DeepCode
 Author: DeepCodeTeam
@@ -24,6 +24,7 @@ Requires-Dist: aiohttp>=3.8.0
 Requires-Dist: anthropic
 Requires-Dist: asyncio-mqtt
 Requires-Dist: docling
+Requires-Dist: google-genai
 Requires-Dist: mcp-agent
 Requires-Dist: mcp-server-git
 Requires-Dist: nest_asyncio
@@ -587,6 +588,14 @@ curl -O https://raw.githubusercontent.com/HKUDS/DeepCode/main/mcp_agent.secrets.
 # Edit mcp_agent.secrets.yaml with your API keys and base_url:
 # - openai: api_key, base_url (for OpenAI/custom endpoints)
 # - anthropic: api_key (for Claude models)
+# - google: api_key (for Gemini models)
+
+# 🤖 Select your preferred LLM provider (optional)
+# Edit mcp_agent.config.yaml to choose your LLM (line ~106):
+# - llm_provider: "google"    # Use Google Gemini models
+# - llm_provider: "anthropic" # Use Anthropic Claude models
+# - llm_provider: "openai"    # Use OpenAI/compatible models
+# Note: If not set or unavailable, will automatically fallback to first available provider
 
 # 🔑 Configure search API keys for web search (optional)
 # Edit mcp_agent.config.yaml to set your API keys:
@@ -623,6 +632,14 @@ uv pip install -r requirements.txt
 # Edit mcp_agent.secrets.yaml with your API keys and base_url:
 # - openai: api_key, base_url (for OpenAI/custom endpoints)
 # - anthropic: api_key (for Claude models)
+# - google: api_key (for Gemini models)
+
+# 🤖 Select your preferred LLM provider (optional)
+# Edit mcp_agent.config.yaml to choose your LLM (line ~106):
+# - llm_provider: "google"    # Use Google Gemini models
+# - llm_provider: "anthropic" # Use Anthropic Claude models
+# - llm_provider: "openai"    # Use OpenAI/compatible models
+# Note: If not set or unavailable, will automatically fallback to first available provider
 
 # 🔑 Configure search API keys for web search (optional)
 # Edit mcp_agent.config.yaml to set your API keys:
@@ -649,6 +666,14 @@ pip install -r requirements.txt
 # Edit mcp_agent.secrets.yaml with your API keys and base_url:
 # - openai: api_key, base_url (for OpenAI/custom endpoints)
 # - anthropic: api_key (for Claude models)
+# - google: api_key (for Gemini models)
+
+# 🤖 Select your preferred LLM provider (optional)
+# Edit mcp_agent.config.yaml to choose your LLM (line ~106):
+# - llm_provider: "google"    # Use Google Gemini models
+# - llm_provider: "anthropic" # Use Anthropic Claude models
+# - llm_provider: "openai"    # Use OpenAI/compatible models
+# Note: If not set or unavailable, will automatically fallback to first available provider
 
 # 🔑 Configure search API keys for web search (optional)
 # Edit mcp_agent.config.yaml to set your API keys:

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8}/deepcode_hku.egg-info/requires.txt

@@ -3,6 +3,7 @@ aiohttp>=3.8.0
 anthropic
 asyncio-mqtt
 docling
+google-genai
 mcp-agent
 mcp-server-git
 nest_asyncio

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8}/mcp_agent.config.yaml

@@ -26,32 +26,32 @@ mcp:
         PYTHONPATH: .
     brave:
       # macos and linux should use this
-      # args:
-      # - -y
-      # - '@modelcontextprotocol/server-brave-search'
-      # command: npx
+      args:
+      - -y
+      - '@modelcontextprotocol/server-brave-search'
+      command: npx
 
       # windows should use this
-      args:
-      # please use the correct path for your system
-      - C:/Users/LEGION/AppData/Roaming/npm/node_modules/@modelcontextprotocol/server-brave-search/dist/index.js
-      command: node
+      # args:
+      # # please use the correct path for your system
+      # - C:/Users/LEGION/AppData/Roaming/npm/node_modules/@modelcontextprotocol/server-brave-search/dist/index.js
+      # command: node
       env:
         BRAVE_API_KEY: ''
     filesystem:
       # macos and linux should use this
-      # args:
-      # - -y
-      # - '@modelcontextprotocol/server-filesystem'
-      # - .
-      # command: npx
-
-      # windows should use this
       args:
-      # please use the correct path for your system
-      - C:/Users/LEGION/AppData/Roaming/npm/node_modules/@modelcontextprotocol/server-filesystem/dist/index.js
+      - -y
+      - '@modelcontextprotocol/server-filesystem'
       - .
-      command: node
+      command: npx
+
+      # windows should use this
+      # args:
+      # # please use the correct path for your system
+      # - C:/Users/LEGION/AppData/Roaming/npm/node_modules/@modelcontextprotocol/server-filesystem/dist/index.js
+      # - .
+      # command: node
 
 
     code-implementation:
@@ -100,14 +100,27 @@ mcp:
       command: python
       env:
         PYTHONPATH: .
+# LLM Provider Priority (选择使用哪个LLM / Choose which LLM to use)
+# Options: "anthropic", "google", "openai"
+# If not set or provider unavailable, will fallback to first available provider
+llm_provider: "google"  # 设置为 "google", "anthropic", 或 "openai"
+
 openai:
   base_max_tokens: 40000
   # default_model: google/gemini-2.5-pro
-  # default_model: anthropic/claude-sonnet-4.5
+  default_model: anthropic/claude-sonnet-4.5
   # default_model: openai/gpt-oss-120b
   # default_model: deepseek/deepseek-v3.2-exp
   # default_model: moonshotai/kimi-k2-thinking
-  default_model: anthropic/claude-3.5-sonnet
+  reasoning_effort: low  # Only for thinking models
   max_tokens_policy: adaptive
   retry_max_tokens: 32768
+
+# Configuration for Google AI (Gemini)
+google:
+  default_model: "gemini-3-pro-preview"
+
+anthropic:
+  default_model: "claude-sonnet-4.5"
+
 planning_mode: traditional

deepcode_hku-1.0.8/mcp_agent.secrets.yaml

@@ -0,0 +1,7 @@
+openai:
+  api_key: ""
+  base_url: ""
+anthropic:
+  api_key: ""
+google:
+  api_key: ""

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8}/requirements.txt

@@ -3,6 +3,7 @@ aiohttp>=3.8.0
 anthropic
 asyncio-mqtt
 docling
+google-genai
 mcp-agent
 mcp-server-git
 nest_asyncio

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8}/utils/llm_utils.py

@@ -12,40 +12,77 @@ from typing import Any, Type, Dict, Tuple
 # Import LLM classes
 from mcp_agent.workflows.llm.augmented_llm_anthropic import AnthropicAugmentedLLM
 from mcp_agent.workflows.llm.augmented_llm_openai import OpenAIAugmentedLLM
+from mcp_agent.workflows.llm.augmented_llm_google import GoogleAugmentedLLM
 
 
 def get_preferred_llm_class(config_path: str = "mcp_agent.secrets.yaml") -> Type[Any]:
     """
-    Automatically select the LLM class based on API key availability in configuration.
+    Select the LLM class based on user preference and API key availability.
 
-    Reads from YAML config file and returns AnthropicAugmentedLLM if anthropic.api_key
-    is available, otherwise returns OpenAIAugmentedLLM.
+    Priority:
+    1. Check mcp_agent.config.yaml for llm_provider preference
+    2. Verify the preferred provider has API key
+    3. Fallback to first available provider
 
     Args:
-        config_path: Path to the YAML configuration file
+        config_path: Path to the secrets YAML configuration file
 
     Returns:
         class: The preferred LLM class
     """
     try:
-        # Try to read the configuration file
-        if os.path.exists(config_path):
-            with open(config_path, "r", encoding="utf-8") as f:
-                config = yaml.safe_load(f)
+        # Read API keys from secrets file
+        if not os.path.exists(config_path):
+            print(f"🤖 Config file {config_path} not found, using OpenAIAugmentedLLM")
+            return OpenAIAugmentedLLM
 
-            # Check for anthropic API key in config
-            anthropic_config = config.get("anthropic", {})
-            anthropic_key = anthropic_config.get("api_key", "")
+        with open(config_path, "r", encoding="utf-8") as f:
+            secrets = yaml.safe_load(f)
+
+        # Get API keys
+        anthropic_key = secrets.get("anthropic", {}).get("api_key", "").strip()
+        google_key = secrets.get("google", {}).get("api_key", "").strip()
+        openai_key = secrets.get("openai", {}).get("api_key", "").strip()
+
+        # Read user preference from main config
+        main_config_path = "mcp_agent.config.yaml"
+        preferred_provider = None
+        if os.path.exists(main_config_path):
+            with open(main_config_path, "r", encoding="utf-8") as f:
+                main_config = yaml.safe_load(f)
+                preferred_provider = main_config.get("llm_provider", "").strip().lower()
+
+        # Map of providers to their classes and keys
+        provider_map = {
+            "anthropic": (
+                AnthropicAugmentedLLM,
+                anthropic_key,
+                "AnthropicAugmentedLLM",
+            ),
+            "google": (GoogleAugmentedLLM, google_key, "GoogleAugmentedLLM"),
+            "openai": (OpenAIAugmentedLLM, openai_key, "OpenAIAugmentedLLM"),
+        }
 
-            if anthropic_key and anthropic_key.strip() and not anthropic_key == "":
-                # print("🤖 Using AnthropicAugmentedLLM (Anthropic API key found in config)")
-                return AnthropicAugmentedLLM
+        # Try user's preferred provider first
+        if preferred_provider and preferred_provider in provider_map:
+            llm_class, api_key, class_name = provider_map[preferred_provider]
+            if api_key:
+                print(f"🤖 Using {class_name} (user preference: {preferred_provider})")
+                return llm_class
             else:
-                # print("🤖 Using OpenAIAugmentedLLM (Anthropic API key not configured)")
-                return OpenAIAugmentedLLM
-        else:
-            print(f"🤖 Config file {config_path} not found, using OpenAIAugmentedLLM")
-            return OpenAIAugmentedLLM
+                print(
+                    f"⚠️ Preferred provider '{preferred_provider}' has no API key, checking alternatives..."
+                )
+
+        # Fallback: try providers in order of availability
+        for provider, (llm_class, api_key, class_name) in provider_map.items():
+            if api_key:
+                print(f"🤖 Using {class_name} ({provider} API key found)")
+                return llm_class
+
+        # No API keys found
+        print("⚠️ No API keys configured, falling back to OpenAIAugmentedLLM")
+        return OpenAIAugmentedLLM
 
     except Exception as e:
         print(f"🤖 Error reading config file {config_path}: {e}")
@@ -101,7 +138,7 @@ def get_default_models(config_path: str = "mcp_agent.config.yaml"):
         config_path: Path to the configuration file
 
     Returns:
-        dict: Dictionary with 'anthropic' and 'openai' default models
+        dict: Dictionary with 'anthropic', 'openai', and 'google' default models
     """
     try:
         if os.path.exists(config_path):
@@ -111,20 +148,34 @@ def get_default_models(config_path: str = "mcp_agent.config.yaml"):
             # Handle null values in config sections
             anthropic_config = config.get("anthropic") or {}
             openai_config = config.get("openai") or {}
+            google_config = config.get("google") or {}
 
             anthropic_model = anthropic_config.get(
                 "default_model", "claude-sonnet-4-20250514"
             )
             openai_model = openai_config.get("default_model", "o3-mini")
+            google_model = google_config.get("default_model", "gemini-2.0-flash")
 
-            return {"anthropic": anthropic_model, "openai": openai_model}
+            return {
+                "anthropic": anthropic_model,
+                "openai": openai_model,
+                "google": google_model,
+            }
         else:
             print(f"Config file {config_path} not found, using default models")
-            return {"anthropic": "claude-sonnet-4-20250514", "openai": "o3-mini"}
+            return {
+                "anthropic": "claude-sonnet-4-20250514",
+                "openai": "o3-mini",
+                "google": "gemini-2.0-flash",
+            }
 
     except Exception as e:
         print(f"❌Error reading config file {config_path}: {e}")
-        return {"anthropic": "claude-sonnet-4-20250514", "openai": "o3-mini"}
+        return {
+            "anthropic": "claude-sonnet-4-20250514",
+            "openai": "o3-mini",
+            "google": "gemini-2.0-flash",
+        }
 
 
 def get_document_segmentation_config(

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8}/workflows/agents/memory_agent_concise.py

@@ -63,8 +63,9 @@ class ConciseMemoryAgent:
 
         # Store default models configuration
         self.default_models = default_models or {
-            "anthropic": "anthropic/claude-sonnet-4.5",
-            "openai": "openai/gpt-4o",
+            "anthropic": "claude-sonnet-4-20250514",
+            "openai": "o3-mini",
+            "google": "gemini-2.0-flash",
         }
 
         # Memory state tracking - new logic: trigger after each write_file
@@ -1422,6 +1423,54 @@ class ConciseMemoryAgent:
                 self.logger.warning("OpenAI response is empty or malformed")
                 return {"content": ""}
 
+        elif client_type == "google":
+            from google.genai import types
+
+            # Convert messages to Gemini format
+            system_instruction = "You are an expert code implementation summarizer. Create structured summaries of implemented code files that preserve essential information about functions, dependencies, and implementation approaches."
+
+            gemini_messages = []
+            for msg in summary_messages:
+                role = msg.get("role", "user")
+                content = msg.get("content", "")
+
+                # Convert role names: "assistant" -> "model"
+                if role == "assistant":
+                    role = "model"
+                elif role not in ["user", "model"]:
+                    role = "user"
+
+                gemini_messages.append(
+                    types.Content(role=role, parts=[types.Part.from_text(text=content)])
+                )
+
+            config = types.GenerateContentConfig(
+                max_output_tokens=5000,
+                temperature=0.2,
+                system_instruction=system_instruction,
+            )
+
+            response = await client.aio.models.generate_content(
+                model=self.default_models.get("google", "gemini-2.0-flash"),
+                contents=gemini_messages,
+                config=config,
+            )
+
+            # Extract content from Gemini response
+            content = ""
+            if response and hasattr(response, "candidates") and response.candidates:
+                candidate = response.candidates[0]
+                if hasattr(candidate, "content") and candidate.content:
+                    if hasattr(candidate.content, "parts") and candidate.content.parts:
+                        for part in candidate.content.parts:
+                            if hasattr(part, "text") and part.text:
+                                content += part.text
+
+            if not content:
+                self.logger.warning("Google response is empty or malformed")
+
+            return {"content": content}
+
         else:
             raise ValueError(f"Unsupported client type: {client_type}")
 

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8}/workflows/agents/memory_agent_concise_index.py

@@ -63,8 +63,9 @@ class ConciseMemoryAgent:
 
         # Store default models configuration
         self.default_models = default_models or {
-            "anthropic": "anthropic/claude-sonnet-4.5",
-            "openai": "openai/gpt-4o",
+            "anthropic": "claude-sonnet-4-20250514",
+            "openai": "o3-mini",
+            "google": "gemini-2.0-flash",
         }
 
         # Memory state tracking - new logic: trigger after each write_file
@@ -1422,6 +1423,54 @@ class ConciseMemoryAgent:
                 self.logger.warning("OpenAI response is empty or malformed")
                 return {"content": ""}
 
+        elif client_type == "google":
+            from google.genai import types
+
+            # Convert messages to Gemini format
+            system_instruction = "You are an expert code implementation summarizer. Create structured summaries of implemented code files that preserve essential information about functions, dependencies, and implementation approaches."
+
+            gemini_messages = []
+            for msg in summary_messages:
+                role = msg.get("role", "user")
+                content = msg.get("content", "")
+
+                # Convert role names: "assistant" -> "model"
+                if role == "assistant":
+                    role = "model"
+                elif role not in ["user", "model"]:
+                    role = "user"
+
+                gemini_messages.append(
+                    types.Content(role=role, parts=[types.Part.from_text(text=content)])
+                )
+
+            config = types.GenerateContentConfig(
+                max_output_tokens=5000,
+                temperature=0.2,
+                system_instruction=system_instruction,
+            )
+
+            response = await client.aio.models.generate_content(
+                model=self.default_models.get("google", "gemini-2.0-flash"),
+                contents=gemini_messages,
+                config=config,
+            )
+
+            # Extract content from Gemini response
+            content = ""
+            if response and hasattr(response, "candidates") and response.candidates:
+                candidate = response.candidates[0]
+                if hasattr(candidate, "content") and candidate.content:
+                    if hasattr(candidate.content, "parts") and candidate.content.parts:
+                        for part in candidate.content.parts:
+                            if hasattr(part, "text") and part.text:
+                                content += part.text
+
+            if not content:
+                self.logger.warning("Google response is empty or malformed")
+
+            return {"content": content}
+
         else:
             raise ValueError(f"Unsupported client type: {client_type}")
 

{deepcode_hku-1.0.7 → deepcode_hku-1.0.8}/workflows/agents/memory_agent_concise_multi.py

@@ -70,7 +70,8 @@ class ConciseMemoryAgent:
         # Store default models configuration
         self.default_models = default_models or {
             "anthropic": "claude-sonnet-4-20250514",
-            "openai": "gpt-4o",
+            "openai": "o3-mini",
+            "google": "gemini-2.0-flash",
         }
 
         # Memory state tracking - new logic: trigger after each write_multiple_files
@@ -657,6 +658,54 @@ class ConciseMemoryAgent:
 
             return {"content": response.choices[0].message.content or ""}
 
+        elif client_type == "google":
+            from google.genai import types
+
+            # Convert messages to Gemini format
+            system_instruction = "You are an expert code implementation summarizer. Create structured summaries of implemented code files that preserve essential information about functions, dependencies, and implementation approaches."
+
+            gemini_messages = []
+            for msg in summary_messages:
+                role = msg.get("role", "user")
+                content = msg.get("content", "")
+
+                # Convert role names: "assistant" -> "model"
+                if role == "assistant":
+                    role = "model"
+                elif role not in ["user", "model"]:
+                    role = "user"
+
+                gemini_messages.append(
+                    types.Content(role=role, parts=[types.Part.from_text(text=content)])
+                )
+
+            config = types.GenerateContentConfig(
+                max_output_tokens=8000,  # Increased for multi-file support
+                temperature=0.2,
+                system_instruction=system_instruction,
+            )
+
+            response = await client.aio.models.generate_content(
+                model=self.default_models.get("google", "gemini-2.0-flash"),
+                contents=gemini_messages,
+                config=config,
+            )
+
+            # Extract content from Gemini response
+            content = ""
+            if response and hasattr(response, "candidates") and response.candidates:
+                candidate = response.candidates[0]
+                if hasattr(candidate, "content") and candidate.content:
+                    if hasattr(candidate.content, "parts") and candidate.content.parts:
+                        for part in candidate.content.parts:
+                            if hasattr(part, "text") and part.text:
+                                content += part.text
+
+            if not content:
+                self.logger.warning("Google response is empty or malformed")
+
+            return {"content": content}
+
         else:
             raise ValueError(f"Unsupported client type: {client_type}")