Update app.py

app.py

@@ -19,6 +19,8 @@ from scripts.text_web_browser import (
     VisitTool,
 )
 from scripts.visual_qa import visualizer
+from scripts.frontmatter_tool import FrontmatterGeneratorTool
+from scripts.text_cleaner_tool import TextCleanerTool
 
 from smolagents import (
     CodeAgent,
@@ -183,6 +185,19 @@ class ToolRegistry:
             TextInspectorTool(model, text_limit),
         ]
 
+    @staticmethod
+    def load_document_tools():
+        """
+        Initialize and return document processing, i.e. sanitisation and indexing, tools.
+
+        Returns:
+            List of document tools
+        """
+        return [
+            FrontmatterGeneratorTool(),
+            TextCleanerTool(),
+        ]
+
     @staticmethod
     def load_image_generation_tools():
         """Initialize and return image generation tools."""
@@ -199,39 +214,54 @@ class ToolRegistry:
 
 # ------------------------ Agent Creation and Execution ------------------------
 def create_agent():
-    """Creates a fresh agent instance with properly configured tools."""
-    # Initialize model
-    model = LiteLLMModel(
-        custom_role_conversions=CUSTOM_ROLE_CONVERSIONS,
-        model_id="openrouter/google/gemini-2.0-flash-001",  # currently serving:
-    )  # DEEPSEEK = openrouter/perplexity/r1-1776 <--- boss model
-
-    # Initialize tools
-    text_limit = 30000
-    browser = SimpleTextBrowser(**BROWSER_CONFIG)
-
-    # Collect all tools in a single list
-    web_tools = ToolRegistry.load_web_tools(model, browser, text_limit)
-    image_generator = ToolRegistry.load_image_generation_tools()
-
-    # Combine all tools into a single list (not a tuple)
-    all_tools = [visualizer] + web_tools + [image_generator]
-
-    # Validate tools before creating agent
-    for tool in all_tools:
-        if not isinstance(tool, Tool):
-            raise ValueError(
-                f"Invalid tool type: {type(tool)}. All tools must be instances of Tool class."
-            )
+    """
+    Creates a fresh agent instance with properly configured tools.
+
+    Returns:
+        CodeAgent: Configured agent ready for use
+
+    Raises:
+        ValueError: If tool validation fails
+        RuntimeError: If agent creation fails
+    """
+    try:
+        # Initialize model
+        model = LiteLLMModel(
+            custom_role_conversions=CUSTOM_ROLE_CONVERSIONS,
+            model_id="openrouter/deepseek/deepseek-chat-v3-0324:free",
+        )
 
-    return CodeAgent(
-        model=model,
-        tools=all_tools,  # Pass a single list containing all tools
-        max_steps=10,
-        verbosity_level=1,
-        additional_authorized_imports=AUTHORIZED_IMPORTS,
-        planning_interval=4,
-    )
+        # Initialize tools
+        text_limit = 30000
+        browser = SimpleTextBrowser(**BROWSER_CONFIG)
+
+        # Collect all tools in a single list
+        web_tools = ToolRegistry.load_web_tools(model, browser, text_limit)
+        doc_tools = ToolRegistry.load_document_tools()  # New document tools
+        image_generator = ToolRegistry.load_image_generation_tools()
+
+        # Combine all tools into a single list
+        all_tools = [visualizer] + web_tools + doc_tools + [image_generator]
+
+        # Validate tools before creating agent
+        for tool in all_tools:
+            if not isinstance(tool, Tool):
+                raise ValueError(
+                    f"Invalid tool type: {type(tool)}. "
+                    f"All tools must be instances of Tool class."
+                )
+
+        return CodeAgent(
+            model=model,
+            tools=all_tools,
+            max_steps=12,
+            verbosity_level=2,
+            additional_authorized_imports=AUTHORIZED_IMPORTS,
+            planning_interval=4,
+        )
+    except (ValueError, RuntimeError) as e:
+        print(f"Failed to create agent: {e}")
+        raise RuntimeError(f"Agent creation failed: {e}")
 
 
 def stream_to_gradio(