Skip to main content
The Control Plane MCP Server enables AI assistants and coding tools to interact directly with your Control Plane infrastructure. Using the Model Context Protocol (MCP), you can create, manage, and deploy resources through natural language conversations.

What is MCP?

The Model Context Protocol (MCP) is an open standard that allows AI models to securely connect to external tools and data sources. With the Control Plane MCP Server, AI assistants can:

Manage Resources

Create and configure GVCs, workloads, and more

Deploy Applications

Build images and deploy applications across global locations

Query Infrastructure

Get real-time information about your Control Plane resources

Automate Workflows

Streamline DevOps tasks through conversational AI

Quick Start

1

Create a Service Account

Create a service account with the appropriate permissions for your use case. Generate an API key that will be used for authentication.
For quick setup, add your service account to the superusers group for full access, or the viewers group for read-only access. See Service Account Permissions below for details.
2

Configure Your Client

Add the Control Plane MCP Server to your AI assistant or coding tool. See the setup guides below for tool-specific instructions.
3

Start Building

Begin interacting with Control Plane through natural language commands.

MCP Server Endpoint

url
string
required
https://mcp.cpln.io/mcp

Authentication

The MCP Server requires authentication via a service account token. Include the token in the Authorization header:
Authorization: <your-service-account-token>
Keep your service account token secure. Never commit tokens to version control or share them publicly.

Compatible Tools

The Control Plane MCP Server works with any AI assistant or development tool that supports remote MCP servers. Below are setup guides for popular tools:
Don’t see your tool listed? Any client that supports remote MCP servers can connect to the Control Plane MCP Server using the standard configuration format. Check your tool’s documentation for remote MCP server setup instructions.

Service Account Permissions

The MCP Server respects all Control Plane policies and permissions. Your service account’s capabilities depend on the groups and policies assigned to it.

Built-in Groups

Control Plane provides two built-in groups for common permission patterns:
GroupPermissionsUse Case
viewersview on all resourcesRead-only exploration, querying infrastructure status
superusersmanage on all resourcesFull automation, creating and modifying all resources

Custom Permissions

For more granular control, you can:
  1. Create custom groups with specific members
  2. Define policies that grant targeted permissions
  3. Assign your service account to groups that match your use case
Create dedicated service accounts for different environments (development, staging, production) with appropriate permission scopes. For detailed information on permissions and policies, see the Policy Reference.
Use CaseRecommended Setup
Quick explorationAdd service account to viewers group
Full development accessAdd service account to superusers group
Production automationCreate custom group with specific policies for required resources
Team-specific accessCreate group per team with policies scoped to their GVCs

Best Practices

Start with Context

Always set your org and GVC context at the start of a conversation

Be Specific

Provide specific values for resources (memory, CPU, replicas) when possible

Confirm Destructive Actions

Configure your AI assistant to confirm before executing delete or update operations

Use Tags

Add tags to resources for better organization and filtering
For the best results, structure your prompts with:
  1. Context: Org and GVC
  2. Action: What you want to do
  3. Target: Which resource
  4. Details: Specific configuration values
Using org "my-org" and GVC "production", create a workload called "api-v2" 
with image nginx:latest, 512Mi memory, port 8080, and minimum 2 replicas.

Next Steps