MCIP is designed to be discoverable and self-documenting. This means your AI agent does not need "hardcoded knowledge" about available tools or their parameters. This dynamic discovery mechanism ensures your integration automatically gains new capabilities as MCIP evolves, without requiring any changes to your client code.
Once connected, your agent's first task is discovering what's possible. This is accomplished by calling the tools/list method.
This method returns a comprehensive catalog of all available functionality, complete with descriptions, parameter schemas, and examples. This discovery mechanism is a "contract" that defines exactly how your agent can interact with MCIP.
| Schema Feature | Description |
|---|---|
| JSON Schema | Each tool description includes a JSON Schema that precisely defines its expected parameters and constraints. |
| Client-Side Use | This schema can be used to generate client code, validate inputs before sending them, and provide intelligent autocomplete in development environments. |
| Rich Detail | The schema includes type information, validation rules, default values, and helpful descriptions for each parameter. |
This discovery-first approach is key to future-proofing your integration. As MCIP adds new tools (like new search filters or "entirely new tools for order management"), your client will automatically see them via the discovery endpoint.
By building a generic tool execution framework that can handle any tool discovered through this mechanism, new MCIP capabilities become available to your users without requiring client updates.