Update app.py

app.py

@@ -141,39 +141,51 @@ Respond in STRICT JSON ONLY:
 
 
 # ---------------------------------------------------
-# Gradio UI Function
+# Gradio UI Function (Uses individual fields)
 # ---------------------------------------------------
-def gradio_interface_fn(payload: RobotWatchPayload):
+def gradio_ui_with_fields(
+    hf_token_input: str,
+    robot_id_input: str,
+    image_file: gr.File # Gradio component for file upload
+):
     """
-    This function acts as the entry point for both the Gradio UI and the MCP Server endpoint.
-    Using the Pydantic model ensures a valid JSON schema is exposed.
+    Handles input from individual Gradio components, converts to Pydantic model, 
+    and calls the core logic.
     """
-    # When called via MCP, the input is already a RobotWatchPayload instance.
-    return run_vlm_analysis(payload)
+    if not image_file or not image_file.path:
+        return {"error": "Image file not uploaded."}
+    
+    # Read the file from the path Gradio provides and convert to base64
+    with open(image_file.path, "rb") as f:
+        image_b64_input = base64.b64encode(f.read()).decode()
+
+    # Create the Pydantic model instance manually
+    payload_instance = RobotWatchPayload(
+        hf_token=hf_token_input,
+        robot_id=robot_id_input,
+        image_b64=image_b64_input
+    )
+    
+    # Call the core logic
+    result = run_vlm_analysis(payload_instance)
+    return result
 
 
 app = gr.Interface(
-    fn=gradio_interface_fn, # Use the single entry point function
-    # Corrected input component from gr.JSON() to gr.Json() as per Gradio documentation
-    inputs=gr.Json(label="Input Payload (Pydantic Schema Applied)"), 
+    fn=gradio_ui_with_fields, # Use the multi-input function for the UI
+    inputs=[
+        gr.Textbox(label="Hugging Face Token", lines=1),
+        gr.Textbox(label="Robot ID", lines=1, value="unknown"),
+        gr.File(label="Upload Image (test.jpg)")
+    ],
     outputs=gr.Json(label="Tool Output"),
-    title="Robot MCP Server",
-    description="A MCP Server to describe image obtained from the CV of a robot/webcam.",
+    title="Robot MCP Server (Field Inputs)",
+    description="Interface for the robot VLM analysis using individual fields.",
     api_name="predict"
 )
 
-# ---------------------------------------------------
-# Explicit MCP API Definition
-# ---------------------------------------------------
-# We explicitly add the API using the Pydantic model for schema generation
-app.api.post(
-    "/mcp/tool/robot_watch", # This defines the exact endpoint path for the tool
-    run_vlm_analysis,      # Link it to the Pydantic-typed function
-    inputs=[RobotWatchPayload], # Use the Pydantic model as the explicit input schema
-    outputs=[dict]             # The output type
-)
-
-
 if __name__ == "__main__":
-    # Launch Gradio with mcp_server=True which hooks up the above API
-    app.launch(mcp_server=True)
+    # Note: When using this method, the automatic MCP schema might become invalid 
+    # again because the *function signature* has changed dramatically.
+    # You might *still* need the `mcp==1.8.1` pin in requirements.txt to work.
+    app.launch(mcp_server=True)