The Agent Skills Directory

COMMAND_EXECUTION (HIGH): The skill explicitly instructs the agent to 'run any additional ".sh" scripts in $TERRAFORM_MODULE/tests'. This facilitates arbitrary code execution of scripts that may be present in untrusted project directories being analyzed or tested by the agent.
EXTERNAL_DOWNLOADS (MEDIUM): The skill references and downloads multiple configuration and workflow files from raw.githubusercontent.com/canonical/. As 'canonical' is not on the Trusted GitHub Organizations list, these sources are considered unverified.
PRIVILEGE ESCALATION (HIGH): The VM setup script (scripts/create-charmkeeper-vm.sh) executes sudo snap install concierge --classic. Classic confinement allows the snap to access system resources outside of its sandbox, posing a risk to the virtual environment.
INDIRECT PROMPT INJECTION (LOW): The skill possesses a significant attack surface by ingesting untrusted data from local terraform modules and executing logic based on it.
Ingestion points: Files located in terraform/ and terraform/tests/ folders of the target project.
Boundary markers: None. No instructions are provided to ignore embedded commands or instructions within the processed project files.
Capability inventory: Use of multipass exec to run terraform init, terraform test, and arbitrary .sh scripts.
Sanitization: None. The skill blindly executes scripts found in the directory structure.

charmkeeper-terraform