Terraform MCP Server

An MCP server that provides tools for generating, validating, planning, and applying Terraform configurations.

Features

• generate_terraform: Generate Terraform configuration files

• terraform_init: Initialize Terraform in a project directory

• terraform_validate: Validate Terraform configuration files

• terraform_plan: Generate and show execution plan

• terraform_apply: Apply configuration to create/update infrastructure

• terraform_destroy: Destroy Terraform-managed infrastructure

• terraform_output: Read Terraform output values

• terraform_state_list: List resources in Terraform state

Related MCP server: Terragrunt GCP Tool MCP

Prerequisites

• Python 3.10+

• Terraform CLI installed and in PATH

• Cloud provider credentials configured (AWS, Azure, or GCP)

Installation

cd terraform-mcp-server
pip install -r requirements.txt

Configuration

Add to your Claude Code settings (~/.claude/settings.json):

{
  "mcpServers": {
    "terraform": {
      "command": "python",
      "args": ["/Users/youruser/Documents/Base/DevOps-ClaudeAi/test-cases/SM1/my-first-vpc/terraform-mcp-server/server.py"],
      "env": {
        "TERRAFORM_WORKSPACE": "/path/to/your/terraform/projects"
      }
    }
  }
}

Usage

Once configured, Claude Code can use these tools:

User: "Initialize terraform in my vpc project"
Claude: [calls terraform_init with project_path]

User: "Run terraform plan"
Claude: [calls terraform_plan]

User: "Apply the changes with auto-approve"
Claude: [calls terraform_apply with auto_approve=true]

Environment Variables

• TERRAFORM_WORKSPACE: Base directory for Terraform projects (default: ~/terraform-projects)

Security Notes

• The terraform_apply and terraform_destroy commands can make real infrastructure changes

• Always review plans before applying

• Use auto_approve=false (default) to require manual confirmation

• Ensure proper cloud provider credentials are configured with least-privilege access