How to get fast, easy insights with the Gremlin MCP Server

Chaos Engineering and reliability testing give you visibility into the actual reliability of your services by simulating real-world failure conditions. But what if you could dig into the testing and results data using AI to quickly uncover new insights?

That’s the logic behind the Gremlin MCP Server.

Released as part of Reliability Intelligence, the Gremlin MCP Server allows you to bring your LLM of choice to explore your Gremlin data and find opportunities to get more out of Gremlin.

What is an MCP server?

A Model Context Protocol (MCP) server is used to connect applications to an LLM. The system consists of three core components: the client, the server, and the Gremlin API.

The client is the AI model you interact with, such as ChatGPT or Claude. Most larger organizations will have specific models that have been vetted and approved by their security team, and these will be compatible with the Gremlin MCP server.

The server receives instructions from the LLM client, then performs the requested work against the Gremlin API before returning results. The Gremlin MCP Server is isolated via containerization and can be run on the server host of your choosing. We highly recommend security hardening and regular auditing of the underlying server, as well as complying with any internal security policies.

The last part is the Gremlin API. We’ve designed the server out of the box to carry out non-destructive operations, so you’re able to safely query your data without worrying about causing damage to your systems or Gremlin installation.

Explore your reliability data and uncover new insights

Once your MCP server is deployed, you can interact with it using your LLM. You and your team can use plain language and prompts to query data, getting answers to questions like:

• Which of my services should I test next?
• What reliability management services are available?
• What critical dependencies aren’t covered?
• Are there gaps in test coverage?

Using more complex prompts, you can also create dashboards to explore the data on a larger scale or report reliability data to your organization.

By default, the API is limited specifically to tasks designed for data exploration, reporting, and evaluation. To make control easier on an organization-wide level, specific MCP roles have been created under Role Based Access Control (RBAC) to scope down API keys as needed.

Because the Gremlin MCP server uses the Gremlin API, it’s also easily extensible, and you can add functionality by using the Gremlin API.

How to deploy a Gremlin MCP Server

Gremlin’s MCP server is now available on GitHub. Setting it up is as easy as deploying the MCP server on your host using your LLM of choice as the client, then connecting to Gremlin using your API key.

To use the Gremlin MCP server, you will need:

• A Gremlin account and REST API key.
• An AI or LLM interface that can run MCP servers, such as Claude Desktop.
• Node.js 22 or higher

From there, you’ll need to clone the repository, install dependencies, build the service, and then configure the MCP client. Then you’ll be ready to go! For detailed installation instructions, see the GitHub repository.

Fast, effective reliability data exploration

The MCP server makes it easy and quick to find new ways to improve the reliability of your services. In fact, we proved the effectiveness while developing the MCP server.

At Gremlin, we use our own platform regularly as part of our reliability program. During MCP server beta testing, it helped us uncover bugs that would have created a problem if they’d found their way to production and revealed several more opportunities for improvement.

We’ve always tried to make it easy to do the right thing for reliability. The Gremlin MCP server helps teams quickly and easily explore their data to uncover insights and make their applications more reliable.

Ready to find out more? Check out our product tours, request a demo, or join us for our MCP server webinar with CTO Sam Rossoff.