The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local helm and terraform CLI tools via bash scripts for infrastructure validation. These scripts utilize array-based command invocation and input validation to prevent command injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface where untrusted configuration data (Helm/Terraform files) enters the agent context through file-path arguments in validate-helm.sh and validate-terraform.sh. Boundary markers for this data are absent, but the risk is managed by a capability inventory limited to local validation tools and sanitization through shell-scripting best practices.

cloud-architect