feat: Filter tools based on mode for Socratic teaching

Problem: Socratic mode should guide learning through questions, not give
direct answers. Tools like segmentation and report generation defeat the
purpose by providing complete answers.

Solution: Filter tools based on agent mode
- Assistant mode: All 8 tools available
- Socratic mode: Only 5 teaching-appropriate tools

Socratic Tools (5):
✓ Grounding - "Show me where you see that"
✓ VQA - Ask targeted questions about specific features
✓ Classification - Show probabilities to guide thinking
✓ Generation - Create practice cases
✓ DICOM Processor - Utility only

Excluded from Socratic (3):
✗ Segmentation - Draws answers automatically
✗ Report Generation - Gives complete diagnosis
✗ Web Browser - Not relevant to case analysis

This ensures Socratic mode truly guides discovery vs giving answers.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

medrax/agent/agent.py

@@ -59,8 +59,23 @@ class Agent:
         self.system_prompt = system_prompt
         self.mode = mode
 
+        # Filter tools based on mode
+        # Socratic mode: Only use tools that help guide learning, not give answers
+        if mode == "socratic":
+            socratic_tool_names = {
+                "xray_phrase_grounding",  # Show locations to guide observation
+                "chest_xray_vqa",  # Ask specific questions about findings
+                "torchxrayvision_classifier",  # Show probabilities to guide thinking
+                "chest_xray_generator",  # Generate practice cases
+                "dicom_processor",  # Utility for file handling
+            }
+            filtered_tools = [t for t in tools if t.name in socratic_tool_names]
+            print(f"Socratic mode: Using {len(filtered_tools)} tools (filtered from {len(tools)})")
+        else:
+            filtered_tools = tools
+
         # Create the parallel tool execution node
-        self.tool_node = ToolNode(tools)
+        self.tool_node = ToolNode(filtered_tools)
 
         # Define the agent workflow with parallel tool execution
         workflow = StateGraph(AgentState)
@@ -71,7 +86,7 @@ class Agent:
         workflow.set_entry_point("agent")
 
         self.workflow = workflow.compile(checkpointer=checkpointer)
-        self.model = model.bind_tools(tools)
+        self.model = model.bind_tools(filtered_tools)
 
     def process_request(self, state: AgentState) -> Dict[str, List[AnyMessage]]:
         """