🐍 ☁️ - MCP server for working with HashiCorp Cloud Platform (HCP) Terraform, enabling AI assistants to…
mcp-server-hcp-terraform
以下摘录自该仓库的 README(片段) · 在 GitHub 读完整 README →
MCP server for working with HashiCorp Terraform Cloud/Enterprise API
Overview
This project provides a MCP server that integrates with the HCP Terraform Cloud/Enterprise API. Through MCP, you can access features such as searching for Terraform modules and retrieving module information.
Prerequisites
- Python 3.13+
- HCP Terraform Cloud/Enterprise account
- Access token
Environment Variables
Set the following environment variables before use:
- HCP_TERRAFORM_TOKEN: HCP Terraform access token (required)
- HCP_TERRAFORM_ORG: HCP Terraform organization name (required)
- HCP_TERRAFORM_BASE_URL: HCP Terraform base URL (optional, default: https://app.terraform.io)
Features
This MCP server provides the following features:
Search Private Modules
Use the hcp_terraform_search_private_modules tool to search for modules in the HCP Terraform Private Registry.
**Parameters**:
- query: Search query
- provider (optional): Provider filter (e.g., aws, gcp, azure)
- limit (optional): Maximum number of results (default: 10)
Get Module Details
Use the hcp_terraform_get_module tool to retrieve detailed information about a specific module from the HCP Terraform Registry.
**Parameters**:
- module_name: Name of the module
- provider: Provider (e.g., aws, gcp, azure)
- registry_name (optional): Registry name (private or public, default: private)
- namespace (optional): Module namespace (uses organization name if not specified)
Usage
License
MIT License