The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous templates and instructions for executing management commands via Azure CLI, Azure Developer CLI (AZD), and Terraform to provision and modify cloud resources.
[EXTERNAL_DOWNLOADS]: The verification recipes in references/recipes/*/verify.md utilize curl to perform health checks against user-deployed endpoints. This is a standard verification procedure for infrastructure deployments.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it ingests and processes data from external sources.
Ingestion points: Deployment outputs and resource lists from az resource list, terraform output, and azd show, as well as HTTP responses from health check endpoints via curl (referenced in references/recipes/*/verify.md).
Boundary markers: Not present; the skill does not explicitly instruct the agent to use delimiters or ignore instructions embedded in the external command outputs.
Capability inventory: The skill has the capability to execute administrative shell commands and interact with Azure resource management APIs via MCP tools and CLI wrappers.
Sanitization: No explicit sanitization or filtering of external data is defined before it is processed by the agent.
[SAFE]: The skill implements a robust safety framework in references/global-rules.md, which mandates the use of ask_user for any action classified as destructive, irreversible, cost-impacting, or security-sensitive (e.g., azd down, az group delete, terraform destroy).

azure-deploy