MCP Server
CloudCasa provides an MCP (Model Context Protocol) server. MCP is an open standard that lets AI assistants and tools, such as Claude Code, connect to external services. The CloudCasa MCP server allows an AI assistant to work with your CloudCasa resources using natural language, for example to check the status of clusters or review recent backup jobs.
The MCP server invokes the CloudCasa APIs on your behalf and authenticates with a CloudCasa API key. It can only do what the role assigned to that API key allows.
Note
The MCP server currently supports read operations only. An AI assistant can view information such as clusters, jobs, and recovery points, but cannot run, create, edit, or delete anything in CloudCasa.
The CloudCasa MCP server is available at:
https://mcp.cloudcasa.io
Connecting an AI assistant
Any AI assistant or tool that supports remote MCP servers over HTTP and allows custom HTTP headers can connect to CloudCasa. Each user connects with their own API key.
In the CloudCasa UI, create an API key by visiting the page Configuration => API Keys. Make sure you select the correct role. For more information, see API Keys.
Configure your AI assistant with the MCP server URL (
https://mcp.cloudcasa.io) and pass the API key in theX-CloudCasa-TokenHTTP header.
Note
Treat API keys like passwords. Anyone with the key can access CloudCasa with the permissions of its role. Consider creating a dedicated API key with a limited role, such as the builtin READONLY role.
Example: Claude Code
To add CloudCasa as an MCP server in Claude Code, first save your API key in an environment variable:
export CC_API_KEY=<YOUR-API-KEY>
Then run:
claude mcp add --transport http cloudcasa https://mcp.cloudcasa.io --header "X-CloudCasa-Token: $CC_API_KEY"
To confirm that the server was added and is connected, run:
claude mcp list
You should see “cloudcasa” listed with a “Connected” status.
See also
For more information, see the Claude Code MCP documentation.
Other MCP clients
Most other MCP clients are configured with a JSON file. The exact format varies by client, but it typically looks like this:
{
"mcpServers": {
"cloudcasa": {
"type": "http",
"url": "https://mcp.cloudcasa.io",
"headers": {
"X-CloudCasa-Token": "<YOUR-API-KEY>"
}
}
}
}
Usage examples
Once connected, you can ask your AI assistant questions about CloudCasa in plain language. The assistant decides which CloudCasa operations to use to answer. For example:
“List all clusters in CloudCasa and show their status.”
“Did any backup jobs fail in the last 24 hours? If so, why?”
“Which backup definitions are configured for cluster my-test-cluster, and when did they last run?”
“Show me the recovery points available for the prod-db namespace.”
“Which object stores are configured, and which backups use them?”
“List any recent alerts.”
Available tools
The MCP server provides the following tools to AI assistants. You don’t normally call these directly; the assistant selects the appropriate tools based on your request.
Tool |
Description |
|---|---|
|
List Kubernetes clusters |
|
Get cluster by ID |
|
List namespaces inventoried in a cluster |
|
List backup policies |
|
Get backup policy by ID |
|
List snapshot backup definitions (from /kubebackups, without copydef or migrationdef) |
|
Get snapshot backup definition by ID |
|
List copy (offload) backup definitions (from /kubeoffloads, with backupdef) |
|
Get copy (offload) backup definition by ID |
|
List recovery points (individual backup runs from /backupinstances; filter state=READY for restorable) |
|
Get recovery point by ID |
|
List restore definitions (configurations, not individual runs) |
|
Get restore definition by ID |
|
List jobs — run-time instances of backup, restore, copy, migration, and DR operations |
|
Get job by ID |
|
List CloudCasa alerts (per-user event records; filter on type, not state) |
|
Get alert by ID |
|
List object stores (S3, Azure, or filestore backup targets) |
|
Get object store by ID |
|
List cloud accounts (AWS, Azure, GCP; state is status.state) |
|
Get cloud account by ID (supports ?embedded={“aws.awsaccount”:1}) |
Note
When referencing CloudCasa resources, always use the _id field value (a 24-character hex string), not the display name.
Revoking access
To revoke an AI assistant’s access to CloudCasa, delete the API key it uses from the Configuration => API Keys page.