The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill identifies a surface for indirect prompt injection due to its core functionality of processing external configuration data using powerful system tools.\n
Ingestion points: Reads and modifies user-provided files including GitHub Actions workflows, Dockerfiles, Kubernetes manifests, and Terraform scripts using the Read tool.\n
Boundary markers: The instructions do not define delimiters or specific 'ignore' directives to prevent the agent from executing instructions that may be embedded within the files it is tasked to manage.\n
Capability inventory: Granted high-privilege access to system tools including kubectl, terraform, helm, docker, and the gh CLI via the Bash tool, alongside Write and Edit file operations.\n
Sanitization: No mechanism is provided to sanitize or validate that the content of the ingested files does not contain adversarial instructions intended to manipulate the agent's behavior.

devops-automator