Skip to main content
The OnboardMe MCP server lets an AI assistant work with a practice’s OnboardMe data. Users can ask the assistant to find a client, check an onboarding, update a proposal, or summarise debtors, and the assistant uses OnboardMe tools to do it. The server uses the Model Context Protocol over Streamable HTTP. Every tool calls the Partner API, so the MCP server never has more access than the API. Practice data is handled the same way as it is for any other API integration.

​
Before you connect

1

Find the practice's region

Each practice is hosted in one region, and each region has its own MCP server. See Regions.
2

Choose how to sign in

Most assistants use Sign in with OnboardMe (OAuth), where a staff member signs in and chooses Read-only or Read & write access. Server-to-server tools can use a practice API key instead. See Authentication and permissions.
3

Add the server to your assistant

Follow the steps for your client in Connect a client.

​
Server URLs

Use the server for the practice’s region. A server in another region cannot sign you in to the practice or see its data.

​
What the assistant can do

The server has tools for clients, contacts, leads, deals, onboardings, proposals, eForms, identity verification and AML, bills, payments, statements, debtors, recurring billing, users and webhooks. See Tools for the full list.

Regions

Which server to use, and how regions affect sign-in.

Connect a client

Claude, ChatGPT, Claude Code and Cursor.

Authentication and permissions

OAuth, API keys, scopes and roles.

Tools

Every tool, grouped by area.