feat : now managing image query tool.

agent.py

@@ -22,68 +22,10 @@
 
 import os
 
-from smolagents import CodeAgent, InferenceClientModel, Tool
-from tavily import TavilyClient
+from smolagents import CodeAgent, InferenceClientModel
 
-# Define a custom tool for Tavily search
-from smolagents import Tool
-from tavily import TavilyClient
-
-tavily_client = TavilyClient(api_key=os.getenv("TAVILY_API_KEY"))
-
-class TavilySearchTool(Tool):
-    name = "tavily_search"
-    description = "Search the web using Tavily."
-    inputs = {
-        "query": {
-            "type": "string",
-            "description": "The search query string.",
-        }
-    }
-    output_type = "string"
-
-    def forward(self, query: str):
-        response = tavily_client.search(query)
-        return response
-    
-class TavilyExtractTool(Tool):
-    name = "tavily_extract"
-    description = "Extract information from web pages using Tavily."
-    inputs = {
-        "url": {
-            "type": "string",
-            "description": "The URL of the web page to extract information from.",
-        }
-    }
-    output_type = "string"
-
-    def forward(self, url: str):
-        response = tavily_client.extract(url)
-        return response
-
-class TavilyImageURLSearchTool(Tool):
-    name = "tavily_image_search"
-    description = "Search for most relevant image URL on the web using Tavily."
-    inputs = {
-        "query": {
-            "type": "string",
-            "description": "The search query string.",
-        }
-    }
-    output_type = "string"
-
-    def forward(self, query: str):
-        response = tavily_client.search(query, include_images=True)
-        
-        images = response.get("images", [])
-        
-        if images:
-            # Return the URL of the first image
-            first_image = images[0]
-            if isinstance(first_image, dict):
-                return first_image.get("url")
-            return first_image
-        return "none"
+from other_tools import ImageQueryTool
+from web_tools import TavilySearchTool, TavilyExtractTool, TavilyImageURLSearchTool
 
 class SmolAlbert(CodeAgent):
     """
@@ -93,6 +35,7 @@ class SmolAlbert(CodeAgent):
     #model_id = "Qwen/Qwen3-Coder-30B-A3B-Instruct"
     #model_id = "Qwen/Qwen3-30B-A3B-Thinking-2507"
     model_id = "Qwen/Qwen3-235B-A22B-Instruct-2507"
+    #model_id = "google/gemma-3-27b-it"
     provider = "auto"
 
     def __init__(self):
@@ -103,12 +46,13 @@ class SmolAlbert(CodeAgent):
         search_tool = TavilySearchTool()
         image_search_tool = TavilyImageURLSearchTool()
         extract_tool = TavilyExtractTool()
+        image_query_tool = ImageQueryTool()
         model = InferenceClientModel(
             model_id=self.model_id,
             provider=self.provider,
             token=os.getenv("HF_API_KEY"))
         self.agent = CodeAgent(
-            tools=[search_tool, image_search_tool, extract_tool], 
+            tools=[search_tool, image_search_tool, extract_tool, image_query_tool], 
             model=model,
             stream_outputs=True,
             instructions=(

agent_ui.py

@@ -333,8 +333,14 @@ class AgentUI:
                     # Detect final answer messages and append to quiet
                     # HACK : FinalAnswerStep messages are produced by _process_final_answer_step and use "**Final answer:**" text
                     if "final answer" in content_text.lower():
-                        # Replace pending with final answer in Quiet
-                        final_msg = gr.ChatMessage(role=MessageRole.ASSISTANT, content=content_text, metadata={"status": "done"})
+                        # Remove everything before and including the "Final answer:" label (and any leading/trailing whitespace/newlines)
+                        answer_only = re.sub(
+                            r"(?s)^.*?\*\*Final answer:\*\*\s*[\n]*",  # (?s) allows . to match newlines
+                            "",
+                            content_text,
+                            flags=re.IGNORECASE,
+                        )
+                        final_msg = gr.ChatMessage(role=MessageRole.ASSISTANT, content=answer_only, metadata={"status": "done"})
                         if quiet_pending_idx is not None:
                             quiet_messages[quiet_pending_idx] = final_msg
                             quiet_pending_idx = None

other_tools.py

@@ -0,0 +1,82 @@
+# The MIT License
+
+# Copyright (c) 2025 Albert Murienne
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+# THE SOFTWARE.
+
+import os
+from pyexpat.errors import messages
+
+from smolagents import InferenceClientModel, Tool
+
+class ImageQueryTool(Tool):
+    """
+    A tool to ask a question about an image given its URL.
+    """
+    name = "image_query"
+    description = "Ask a question about an image given its URL."
+    inputs = {
+        "image_url": {
+            "type": "string",
+            "description": "The URL of the image to analyze.",
+        },
+        "question": {
+            "type": "string",
+            "description": "The question to ask about the image.",
+        }
+    }
+    output_type = "string"
+
+    def __init__(self):
+        """
+        Construct the ImageQueryTool with a specific model.
+        """
+        # call superclass constructor
+        super().__init__()
+        # Initialize the model
+        self.model = InferenceClientModel(
+            model_id="google/gemma-3-27b-it",
+            provider="auto",
+            token=os.getenv("HF_API_KEY")
+        )
+
+    def forward(self, image_url: str, question: str):
+        """
+        Forward method to process the image URL and question.
+        """
+        messages = [
+            {
+                "role": "user",
+                "content": [
+                    {"type": "text", "text": question},
+                    {"type": "image_url", "image_url": {"url": image_url}}
+                ],
+            }
+        ]
+
+        res = self.model(messages)
+        return res.content
+
+if __name__ == "__main__":
+    tool = ImageQueryTool()
+    response = tool.forward(
+        image_url="https://upload.wikimedia.org/wikipedia/commons/9/99/Black_square.jpg",
+        question="What is this?"
+    )
+    print("Response:", response)

web_tools.py

@@ -0,0 +1,91 @@
+# The MIT License
+
+# Copyright (c) 2025 Albert Murienne
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+# THE SOFTWARE.
+
+import os
+
+from smolagents import Tool
+from tavily import TavilyClient
+
+tavily_client = TavilyClient(api_key=os.getenv("TAVILY_API_KEY"))
+
+class TavilySearchTool(Tool):
+    """
+    A tool to perform web searches using the Tavily API.
+    """
+    name = "tavily_search"
+    description = "Search the web using Tavily."
+    inputs = {
+        "query": {
+            "type": "string",
+            "description": "The search query string.",
+        }
+    }
+    output_type = "string"
+
+    def forward(self, query: str):
+        response = tavily_client.search(query)
+        return response
+    
+class TavilyExtractTool(Tool):
+    """
+    A tool to extract information from web pages using the Tavily API.
+    """
+    name = "tavily_extract"
+    description = "Extract information from web pages using Tavily."
+    inputs = {
+        "url": {
+            "type": "string",
+            "description": "The URL of the web page to extract information from.",
+        }
+    }
+    output_type = "string"
+
+    def forward(self, url: str):
+        response = tavily_client.extract(url)
+        return response
+
+class TavilyImageURLSearchTool(Tool):
+    """
+    A tool to search for image URLs using the Tavily API.
+    """
+    name = "tavily_image_search"
+    description = "Search for most relevant image URL on the web using Tavily."
+    inputs = {
+        "query": {
+            "type": "string",
+            "description": "The search query string.",
+        }
+    }
+    output_type = "string"
+
+    def forward(self, query: str):
+        response = tavily_client.search(query, include_images=True)
+        
+        images = response.get("images", [])
+        
+        if images:
+            # Return the URL of the first image
+            first_image = images[0]
+            if isinstance(first_image, dict):
+                return first_image.get("url")
+            return first_image
+        return "none"