The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes high-privileged CLI tools including az deployment and terraform apply to provision cloud infrastructure. This functionality is the primary intended purpose of the skill and is protected by mandatory status gates (meta.status === 'approved') and explicit user confirmation steps for destructive actions.- [DYNAMIC_EXECUTION]: The skill dynamically generates Bicep and Terraform configuration files in the infra/ directory based on infrastructure plans. These generated scripts are subsequently executed by the agent. The generation logic incorporates official Azure resource schemas and naming constraints fetched at runtime to ensure compliance and validity.- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from user-provided workload descriptions and external documentation fetched via microsoft_docs_fetch. This data influences the generation of infrastructure code.
Ingestion points: User-provided workload descriptions and documentation content fetched from Microsoft Learn.
Boundary markers: None explicitly defined for the sub-agent summarization process.
Capability inventory: File system writes to the infra/ and .azure/ directories, and execution of az and terraform CLI tools.
Sanitization: The skill implements automated syntax validation (az bicep build, terraform validate) and a comprehensive verification phase to check for resource compatibility and dependency conflicts before execution.- [EXTERNAL_DOWNLOADS]: The skill utilizes MCP tools to fetch documentation and best practices from official Microsoft domains (learn.microsoft.com). Additionally, terraform init facilitates the download of providers from official registries. These operations target well-known and trusted technology services.

azure-enterprise-infra-planner