Hi, I am trying out the simple-chatbot example examples/clients/simple-chatbot/mcp_simple_chatbot . I have created approx 4 servers and included them in server_config.json. Each server has approx 2 tools and they are all passed to the model as suggested in lines 407-417 of main.py.
My prompt for system message is:
"You are a helpful assistant with access to these tools:\n\n"
f"{tools_description}\n"
"If a tool is needed, choose the appropriate tool based on the user's question. The tool can come ONLY from the list above."
"If no tool is needed, reply directly.\n"
"IMPORTANT: When you need to use a tool, you must respond with "
"the exact JSON object format below first:\n"
"{\n"
' "tool": "tool-name",\n'
' "arguments": {\n'
' "argument-name": "value"\n'
" }\n"
"}\n\n and then always use the call_tool function for the tool.\n"
"If you need to provide additional information to the tool, please ask the user to provide it.\n"
"After receiving a tool's response:\n"
"1. Transform the raw data into a natural, conversational response\n"
"2. Keep responses concise but informative\n"
"3. Focus only on the most relevant information\n"
"4. Use appropriate context from the user's question\n"
"5. Avoid simply repeating the raw data\n"
"6. When asked information about specific tools, use ONLY information provided by existing tools\n"
"7. If not all required information to run the selected tool are provided ask the user to add them\n"
f"Please use only the tools that are explicitly defined in {tools_description}.\n\n"
I do see that the tools, their description and parameters are passed correctly BUT 1.the model does not print the tool it chooses to use every time AND it sometimes runs the tools I want or other times (depending if I have asked it something else first or not) it just tells me instructions on how to run the tools (sometimes hallucinating). could this be a problem of the model I am using, the prompt or the set up of the example?