The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches Azure resource specifications and schemas from official and trusted sources, including the Azure organization on GitHub (Azure/bicep-types-az and Azure/Azure-Verified-Modules). It also provides instructions for manually installing the tfpluginschema utility from a community repository to assist with provider schema queries.
[COMMAND_EXECUTION]: Development tasks are performed using a local ./avm wrapper script and standard command-line tools such as git, terraform, and powershell. The Get-AzureSchema.ps1 script is used to automate the retrieval of resource definitions for development purposes.
[DATA_EXFILTRATION]: No patterns of sensitive data exfiltration were detected. The skill follows standard practices for handling Azure subscription IDs via environment variables and does not attempt to access or transmit credentials like SSH keys or environment secrets to external domains.
[PROMPT_INJECTION]: The instructions use strong prescriptive language (e.g., "MUST", "SHOULD") to enforce AVM coding standards. These are functional requirements for the development workflow and do not attempt to bypass AI safety filters or override system instructions.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface as it processes Terraform files and external JSON schemas. However, the risk is mitigated by the use of structured data formats (JSON) and standard Terraform validation tools. The severity is low and consistent with development-focused skills.

avm-terraform-module-development