NoVo’s MCP server gives an agent two ways to answer a market question. It can call a data tool and work with the numbers. Or it can call ask_novo, which puts a plain-English question to Dr. NoVo, the Financial Markets Super Intelligence, and returns his answer. Knowing which to reach for saves time and avoids a common muddle.
What the data tools are for
Data tools return structured readings. get_dealer_levels returns levels. get_volatility_record returns percentiles. get_stocks_onchain returns perps and stock tokens. Each field has a name, a value and a time. Use them when you need a figure you can quote, compare or store.
They also suit anything your agent will compute on. If you want the distance from spot to the call wall, the agent needs both numbers as numbers. A paragraph of prose is the wrong input for arithmetic.
What ask_novo is for
ask_novo is not a data endpoint. It is for interpretation. You ask what the picture means, how two readings relate, or what a term implies in today’s conditions. Dr. NoVo answers in sentences and names what he used. That naming is what lets you check him, the same standard set out in an answer you cannot check is an opinion.
The muddle to avoid
The muddle is asking for a number through the interpretive tool, then having your own model paraphrase the reply. Now the figure has passed through two rounds of prose. Each round can round it, drop its time or soften a caveat. For a number, go to the data tool. NoVo’s own tool notes say the same: use ask_novo for interpretation and the others for numbers.
Using both
The two work well in sequence. Have the agent fetch the readings first. Then put the question to ask_novo. You end up with the raw figures on one side and a markets SI’s read on the other. If the read mentions a level, you can hold it against the payload.
This is also a fair division of labor. Your general model is good at running the conversation and choosing tools. It was not built around dealer positioning. Dr. NoVo was. What that difference amounts to is the subject of what Dr. NoVo adds over raw data.
Cost and pace
A data call is light and quick. A read takes longer, because there is reasoning behind it. On the paid plans ask_novo has its own daily allowance, counted apart from the data tools. So spend it on questions that need thought. Do not use it to look up a number a data tool returns directly.
What neither will do
Neither kind of tool predicts and neither advises. A data tool returns where things stand. Dr. NoVo explains what the readings mean together. Ask either one what to buy and the honest answer is that this is a readout. Both live on the MCP & API, and Dr. NoVo can be asked directly on the site.