We’ve launched Merge MCP to help AI companies leverage our integrations in minutes! Here’s how to use it

The Model Context Protocol (MCP) is quickly becoming the go-to standard for facilitating communication between LLMs and outside data sources.

The reasons behind its early success aren’t hard to discern: It’s easy to use, its open-source model has helped make it widely available, and it’s been adopted by leading AI companies (e.g., OpenAI)—prompting the broader AI market to trust it. 

We’re committed to helping AI companies build the most reliable and powerful integrations at scale, so we’ve released Merge MCP to help any company use the Model Context Protocol across our integrations and endpoints!

Here’s more on how Merge MCP works, how you can use it, and what you can do to access it.

{{this-blog-only-cta}}

Overview on Merge MCP 

We’ve built an MCP server that allows any LLM to access our integrations’ endpoints as tools. 

Our integrations will also operate the same through MCP. 

Once your user authenticates a connection with Merge MCP, we'll handle the integration’s rate limits, pagination, day-to-day maintenance, and more; we’ll also continue to provide the enterprise-grade security features you need, such as access control levels (ACLs) and Common Model Scopes; and your customer-facing teams can still use our Integration Observability features to manage the integrations.

We’ve even made Merge MCP’s tools highly detailed to help you review the available tools and enable your LLM to choose the appropriate tools consistently and in real-time. 

Best of all, the process of adopting Merge MCP is incredibly simple. You’ll just need to write a few lines of code (you can find them near the end of this article) to add Merge MCP as a set of tools for your LLM. From there, your LLM can read and write data across our 220+ integrations!

https://www.merge.dev/blog/api-vs-mcp?blog-related=image

Example of using Merge MCP

Since Merge MCP lets an LLM perform a certain task on integrated customer data based on an end user’s prompt, the use cases for Merge MCP are endless.

To make it more tangible, let’s walk through how Merge MCP can support a specific ticketing use case:

1. One of your users enters a prompt that details a specific ticket they want to create.

2. The LLM decides on the tool to use. In this case, it would pick the create_ticket tool to add the ticket in the customer's integrated project management system.

3. The LLM can then confirm that the ticket has been created and even share a link to it in case the user wants to review it.

Related: Best practices for using the Model Context Protocol

How to access Merge MCP

Merge MCP is now available to all of our customers. We’ve also made all of our Common Model endpoints available as tools in Merge MCP so that it can support any of your product’s AI use cases!

To get started with Merge MCP, you’ll just need to write a few lines of code: 

{
  "mcpServers": {
    "merge-mcp": {
      "command": "uvx",
      "args": ["merge-mcp"],
      "env": {
        "MERGE_API_KEY": "xxxxxxx",
        "MERGE_ACCOUNT_TOKEN": "xxxxxxx"
      }
    }
  }
}

To learn more about Merge MCP, you can visit our Docs page. We also encourage you to either schedule a demo with one of our integration experts or (if you already use Merge) reach out to your dedicated customer success manager.

We can’t wait to see all of the innovative ways that our customers use Merge MCP!