The Agent Skills Directory

Command Execution: The skill utilizes various scripts to perform deployment readiness checks. These include validate-deployment.sh for Bicep templates, validate-terraform.sh for Terraform plans, and workflow.sh to manage the overall validation sequence. These scripts invoke tools such as the Azure CLI (az), Terraform, and Docker to verify environment configuration and infrastructure code.
Indirect Prompt Injection Surface: The validation logic involves reading and analyzing project files from the local workspace, such as azure.yaml, Dockerfile, and C# source code (.cs). While necessary for the skill's function, this processes external data that could influence the agent's behavior.
Ingestion points: Project-specific files including azure.yaml, Dockerfile, main.tf, main.bicep, and C# source files are read by the validation scripts located in references/scripts/ and references/recipes/.
Boundary markers: The skill does not use specific delimiters or instructions to ignore potential instructions embedded within the analyzed project files.
Capability inventory: The skill has the capability to run infrastructure tools including az, terraform, azd, and docker, and it manages its state in .azure/validate-status.json.
Sanitization: Content from the analyzed files is inspected using standard utilities like grep and sed to identify configuration patterns without explicit sanitization steps before reporting results to the user.

azure-validate