The Agent Skills Directory

Sensitive Infrastructure Data Access: The skill involves commands like terraform show -json and terraform state pull to analyze infrastructure state. As Terraform state files frequently contain sensitive information (such as credentials or private network details) in plaintext, the skill correctly includes a cautionary note instructing the agent not to output or log these contents.
External Resource References: The skill references external HCL style guides and module specifications hosted on HashiCorp's official GitHub repository. These references are used to provide the agent with additional context on community standards and are sourced from a well-known, trusted organization.
Command Line Interaction: The skill utilizes standard command-line tools, including terraform and jq, to perform logical refactoring tasks. These operations are restricted to infrastructure management within the provided source_directory.
Indirect Prompt Injection Surface: This skill processes user-provided Terraform source code and configuration files. While this is a standard requirement for refactoring, there is a theoretical surface where malicious instructions could be embedded in the source configuration being analyzed. The skill's focus on structured HCL transformation and explicit state management minimizes the risk of the agent executing unintended instructions found in the data.

refactor-module