The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses bash scripts to execute standard cloud infrastructure CLI tools.
validate-helm.sh executes helm lint and helm template to verify Kubernetes manifests.
validate-terraform.sh executes terraform init, terraform validate, and terraform fmt to verify infrastructure-as-code.
These commands are consistent with the skill's primary purpose and use secure array-based argument handling in the shell scripts to prevent shell injection.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes and displays the contents of user-provided Helm charts and Terraform files.
Ingestion points: The skill reads directory paths and file contents via script arguments (CHART_PATH, TF_DIR).
Boundary markers: There are no explicit delimiters or instructions to the LLM to ignore embedded content in the script output.
Capability inventory: The skill has the ability to execute subprocesses and read file outputs.
Sanitization: While the scripts prevent shell injection, the output of rendered templates is returned directly to the agent's context without sanitization, which could contain instructions intended to influence the LLM.

cloud-architect