The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes make lint and make validate-kustomize, which execute logic defined in the local Makefile. This represents a dynamic execution surface where malicious content in the repository's own build files could lead to unauthorized command execution.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes external data without sanitization or boundary markers.
Ingestion points: Untrusted data enters the agent context via the user-supplied $ARGUMENTS and through the contents of kustomization.yaml and the output of OpenShift metadata commands (oc get csv).
Boundary markers: No delimiters or isolation instructions are used to distinguish untrusted data from the agent's internal logic.
Capability inventory: The skill allows the agent to write and edit files, and execute various Bash commands including oc, kustomize, and make.
Sanitization: The skill lacks logic to validate or escape user input before it is interpolated into shell commands (e.g., kustomize build <target-path>/), which could be leveraged for command injection.

add-cnpg-database