Debug mode dynamically per request

[user1500177]

I am using agno 2.5.16 ,is it possible for :

1) Setting the debug mode in agno dynamically for the specific request ? (that is only when we are needin debugging we need to do that is that possible ? )

Using AGUI interface also , is there any way by which we could know of this - it would be only useful when we have to send a request which will enable the debug mode so as to know correctly what is the issue at that time , else always the app would be running in the DEBUG mode

2)Setting the Model as dynamic pre request ?

3)Setting the tools as dynamic per request , but the thing is that i need the tools to be shortlisted by the llm from the wide variety of 128plus tools avalble based on the users intent , and give only those tools (Currently i had a strategy which enabled only to divide the tools into small groups and for varous domain i had NESTED SUBTEAMS UNDER the same parent TEAM and then i used the agents with short listed 20 or so tools for that specific one , BUT - this had a performance hit as 2 times delegation etc , IS there a better IDEA to do this ? ) - i know that agno support the dynamic tools - but is there any idea on the callable i shoudl use for shortlisting the tools properly form alll those tools avaible in the MCP server based on the task of the user or is THE CURRENT approach better ?

If i give the tool_instace created with the MCPTools(url) , i am only able to give it fully now right

server_params = StreamableHTTPClientParams(
url=mcp_url,
headers=headers,
timeout=90.0,
)

mcp_tools = MCPTools(
server_params=server_params,
transport="streamable-http",
timeout_seconds=90,
refresh_connection=True
)

i give this tools to the agent as follows :

agent = Agent(
name="example-agent",
id="example-agent",
role="Agent role",

model=model,

tools=[mcp_tools],
skills=skills,

instructions=[
    "You have a skill available via get_skill_instructions. Load it when needed. , and also other EXPLICIT instructions for the domain knowledge",
],

resolve_in_context=True,
reasoning=enable_agent_reasoning,
tool_call_limit=tool_call_limit,

tool_hooks=[secure_tool_hook],

)

In the tools section i can do something for a callble using - but what should be the logic for it to mkae sure that the correct tools are indeed given form the MCP server that is availble

[user1500177]

Hey , COULD ANY ONE HELP ME ?