This playbook explains how integration partners should work with Keragon MCP.
It helps you determine whether you should build an MCP client or a Keragon connector, clarifies Keragon’s role in the architecture, and shows how your product will appear inside Keragon. It is designed to eliminate confusion and help you choose the correct integration path before starting development.
Who this is for
This playbook is for integration partners evaluating how to work with Keragon MCP and unsure whether they should:
Build their own MCP server
Build an MCP client
Build a Keragon connector
If you are evaluating your integration path, start here.
Note: Keragon acts as an MCP server
All supported integrations fall into one of two categories:
- You connect to Keragon as an MCP client
- You build a Keragon connector to expose your API as tools
Start here: What are you trying to achieve?
Choose the scenario that matches your desired path:
Scenario 3: You want to build your own MCP server so Keragon can connect to it
Scenario 1: You're an AI vendor and want to connect your product to other external systems through Keragon
Example: You offer an AI assistant, agent platform, or automation tool and want your users to access healthcare systems through Keragon.
What this means
You are acting as an MCP client.
Keragon is the MCP server connecting your AI product to other external tools.
What you need to do
Follow the official MCP client onboarding process:
Complete the required questionnaire
Provide our team with a test account so we can confirm the integration works as expected
For the full process please read here.
How you will appear to Keragon users
Keragon users will be able to configure their MCP servers to connect to your MCP client like below.
Scenario 2: You are a software vendor and want your product to be usable by AI tools through Keragon
Example: You provide an EHR, CRM, billing platform, or other healthcare system and want AI agents (e.g. Claude, ChatGPT etc) to call your API through Keragon.
What this means
You need a Keragon connector.
Keragon will expose your connector’s components as MCP tools automatically.
Important
Any connector that exists in Keragon becomes usable:
Inside standard Keragon workflows
As a tool inside Keragon MCP
There is no separate MCP-only integration path.
What you need to do
Build a Keragon connector. You can build a connector self-served by visiting our Partner dashboard
This enables:
Workflow steps
MCP tool exposure
Visibility in Keragon as an integration
How you will appear to Keragon users
Keragon users can either use your connector to choose MCP tools or even as a normal workflow step
Scenario 3: You want to build your own MCP server so Keragon can connect to it
This is not supported. If your product can be connected via an API then probably you should choose Scenario 2 above and build a Keragon connector instead.
Need help?
If you have questions or would like feedback on your page , contact the Keragon partnerships team through your partner dashboard.