Custom MCP Servers
The default DealCloud MCP Server exposes your full site schema — every object and field — to any connected AI client. Custom MCP Servers let admins scope that exposure: defining exactly which objects and fields a given AI client can access. This is useful when different teams, workflows, or client applications need different data surfaces.
Each Custom MCP Server is paired with a Schema Contract, which defines the curated subset of objects and fields that server exposes.
When to Use a Custom Server
- You want to limit what an AI client can see to a specific set of objects or fields
- Different teams or use cases need different data surfaces from the same DealCloud site
- You are connecting multiple AI clients and want each to have a defined, isolated schema pairing
- You want to rename or re-map objects and fields for a specific integration without changing the underlying site structure
Schema Contracts
A Schema Contract is a named, scoped selection of your DealCloud data model — the objects and fields you choose to expose, with the naming and structure you define. It's the handshake between DealCloud and the AI client: instead of exposing the full site schema, a contract gives the client a curated, stable view it can reason over accurately.
Setting Up a Schema Contract
Schema Contracts are found under Tools > Schema Contracts
Requires the Schema Contracts capability enabled in User Management.
Create a New Contract
Click New Contract, give it a name and an optional description, then click Continue. Use a name that reflects the integration or AI client it's intended for.
Select Objects and Fields
You'll see a list of all objects in your DealCloud instance. Check the objects and fields you want to include. Clicking an object or field lets you edit its Name and API Name — this is where you can rename or re-map your schema for the integration without changing the underlying site structure. Add a Description or Internal Note as needed.
When done, click Save and Continue.
Review and Publish
A review screen shows each selected object alongside its fields, with metadata including API names, field types, and whether fields are required or read-only. This is what the connected AI client will see.
When satisfied, click Save and Publish. The contract's status updates to Published and is immediately available to pair with an MCP server.
Configuring a Custom MCP Server
Once a Schema Contract is published, you can create an MCP Server scoped to it.
Create an MCP Server
From Tools > MCP Servers, click New MCP Server. Give it a name and description and select your Schema Contract.
Publish the Server
Click Save and Publish to deploy the server. Once published, copy the MCP endpoint URL and share it with the users or teams who will connect to it.
Each Custom MCP Server exposes only the objects and fields defined in its paired Schema Contract. AI clients connecting to it will only be able to access data within that scope. Individual user permissions are always respected on top.