MCP is the interface to IQ's intellect.
IQ is not just a toolbox. It is a fully cognitive operating system.
MCP for IQ allows you to connect your favourite LLM as a command and control agent to IQ's rich cognitive world.
By leveraging MCP, your LLM can interact with IQ's cognitive capabilities seamlessly, enabling more intelligent and context-aware decision-making.
A protocol should carry world state, not just commands.
The modern AI stack is fragmented. Every tool calls itself a "context layer," but most of them only pass fragments: a document id, a query string, a function name, or a short memory snippet.
That is not enough. An AI needs to know what matters, what is current, what changed, what is true now, and what the model is allowed to infer from the surrounding system.
Today’s problem is not intelligence. It is integration.
Tool APIs are narrow
They expose operations, but they rarely expose the durable structure behind them. An LLM can call a function without understanding your world that function belongs to.
Memory is shallow
Conversation memory often captures the last turn, not the actual entities, events, and relationships that define the situation.
IQ turns MCP into a world model protocol.
MCP should not merely fetch data. It should expose meaning.
In IQ, the protocol can describe current state, historical context, identity, edges, observations, rules, and provenance. That means the model can ask for the right slice of reality instead of guessing from a noisy prompt.
The point is simple: a model cannot act well on a world it cannot inspect structurally.
The right protocol unlocks real reasoning.
Identity
Who or what is this entity, and how is it tracked over time?
Relations
What connects this thing to other things, and why does the connection matter?
State
What is true now, what was once true, and what changed between those moments?
Provenance
Where did this fact come from, and how much confidence should the system place in it?