<a id="mcp"></a>

# ScyllaDB Cloud MCP Server BETA

#### NOTE
ScyllaDB Cloud MCP is currently in Beta. It offers read-only tools to help you get information about your clusters and account. Additional tools will be added in upcoming releases.

## Overview

[Model Context Protocol (MCP)](https://modelcontextprotocol.io/) is an open standard for connecting AI agents to external tools and data sources.

The ScyllaDB Cloud MCP server implements this standard, exposing the [ScyllaDB Cloud API](https://cloud.docs.scylladb.com/stable/api.html) as tools that AI agents can discover and call directly.

It can be used with any AI client that supports MCP. For example:

* [Claude](https://code.claude.com/docs/en/mcp)
* [Cursor](https://docs.cursor.com/context/model-context-protocol)
* [OpenCode](https://opencode.ai/docs/mcp-servers/)

For a ready-to-use agent setup, see the [ScyllaDB agent skills](https://github.com/scylladb/agent-skills) repository.

## ScyllaDB MCP Tools

The following tools are available in the current Beta release. They are all read-only, providing a safe way to query information about your ScyllaDB Cloud environment:

| Tool                     | Description                                                                                                   |
|--------------------------|---------------------------------------------------------------------------------------------------------------|
| `get_account_info`       | Get account details and associated cloud accounts.                                                            |
| `list_clusters`          | List all clusters with optional metrics.                                                                      |
| `get_cluster_overview`   | Get cluster details, data centers, and nodes in a single call.                                                |
| `get_cluster_health`     | Get node health, data center metrics, and cluster-level metrics.                                              |
| `get_cluster_requests`   | List cluster operation requests with optional type and status filters, or get details for a specific request. |
| `track_cluster_progress` | Track the progress of a cluster creation request, including status, progress, and completion state.           |
| `get_documentation`      | Browse API documentation, models, and product documentation.                                                  |

More tools will be added in upcoming releases, including tools for creating clusters and updating scaling policies.

## Connecting to the MCP Server

### Prerequisites

Before connecting, make sure you have the following:

* [A ScyllaDB Cloud cluster](https://cloud.docs.scylladb.com/stable/scylladb-quickstart/index.html)
* An MCP-compatible client
* [Node.js](https://nodejs.org/en/download)

### Configure the MCP Connection

To connect to the ScyllaDB Cloud MCP server you need:

* **MCP Server URL:** `https://mcp.cloud.scylladb.com`
* **Access Token:** A personal API token. See [Create a Personal Token for Authentication](https://cloud.docs.scylladb.com/stable/api-docs/create-api-token.html) for instructions.

The following example shows how to configure the connection in OpenCode. For other MCP clients, refer to their documentation.

1. Sign in to [ScyllaDB Cloud](https://cloud.scylladb.com/account/login) and [create a personal access token](https://cloud.docs.scylladb.com/stable/api-docs/create-api-token.html).
2. Open the OpenCode configuration file `opencode.jsonc`.
3. Add the MCP connection under the `mcp` key. Provide the ScyllaDB Cloud MCP server URL and set the `Authorization` header to your personal token, prefixed with the `Bearer` identifier:
   ```json
   "mcp": {
     "scylladb-cloud": {
       "type": "remote",
       "url": "https://mcp.cloud.scylladb.com",
       "headers": {
         "Authorization": "Bearer <PERSONAL_API_TOKEN>"
       },
       "timeout": 30000
     }
   }
   ```

   Replace `<PERSONAL_API_TOKEN>` with the token you generated in step 1.

Once you are connected to the MCP server, you can list its tools, run them, and inspect their responses.

![Example of listing and running MCP tools in an AI client](api-docs/images/mcp-example.png)
