The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes three Python utilities (crd_validator.py, reconcile_lint.py, and operator_capability_audit.py) that ingest and analyze project files from the user's workspace.
Ingestion points: These scripts read local files (YAML and Go source) specified via command-line arguments such as --crd, --controller, and --operator-dir.
Boundary markers: The tools do not implement delimiters or explicit warnings to the agent regarding instructions that might be embedded in the analyzed content (e.g., within YAML descriptions or Go comments).
Capability inventory: The scripts are restricted to file system read operations (open, os.walk) and string analysis using regex. They do not possess capabilities for network communication, subprocess execution, or dynamic code evaluation.
Sanitization: The tools perform heuristic analysis on raw file content and report findings directly to the console without sanitization or filtering of the input text.

kubernetes-operator