Deploy via MCP (AI Agents)

Deploy infrastructure using AI agents through the Bluebricks Model Context Protocol (MCP) server.

For the full MCP setup guide -- including client configuration for VS Code, Claude Desktop, Cursor, and Windsurf -- see the MCP documentation.

Prerequisites

• Bluebricks CLI installed (v1.54.1 or later)

• Authenticated with Bluebricks (bricks login)

• MCP-compatible AI client configured (see MCP docs)

• Existing blueprints and collections configured

Quick Setup

All MCP clients use the same server configuration:

• Command: bricks

• Args: ["mcp"]

The MCP server uses your existing CLI authentication (bricks login).

Deployment Workflow

1. Discover blueprints -- ask your AI agent what blueprints are available.

2. Request deployment -- describe what you want deployed and where:

• "Deploy the PostgreSQL blueprint to staging"

• "Install the AWS VPC blueprint in production"

3. Review the plan -- the agent shows resource specifications, cost estimates, and security considerations.

4. Approve -- confirm the deployment to proceed.

5. Monitor -- ask for deployment status updates.

Available MCP Tools

• bricks-list-blueprints -- discover infrastructure templates

• bricks-list-environments -- view configured collections

• bricks-list-deployments -- query deployments with filtering

• bricks-install -- deploy blueprints to environments

• bricks-deployment-approve -- approve pending deployments

• bricks-deployment-status -- monitor deployment progress

See also

• MCP documentation -- full setup and configuration guide

• How to Plan Deployments -- CLI-based deployment planning

• How to Authenticate with CLI -- authentication setup