The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/lib/monitoring.py uses subprocess.run to execute the gcloud CLI for authentication. The command is invoked using a list of arguments (e.g., ['gcloud', 'auth', 'application-default', 'print-access-token']), which is a secure implementation that prevents shell injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: The skill communicates with monitoring.googleapis.com to fetch environment metrics. This is the official endpoint for Google Cloud Monitoring, which is a well-known and trusted service.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it is designed to ingest and analyze external data such as system logs.
Ingestion points: System telemetry, environment configuration, and component logs are retrieved from Google Cloud via the gcloud CLI and Monitoring API.
Boundary markers: The skill does not explicitly define delimiters to separate ingested log data from its own instructions in the analysis scripts.
Capability inventory: The skill can execute local Python scripts and perform gcloud operations to inspect environment settings and logs.
Sanitization: The metrics processing scripts handle structured JSON and numerical data; logs are retrieved as YAML/text without specialized sanitization.
Assessment: This surface is inherent to the skill's primary function as a diagnostic tool. Given the use of trusted infrastructure and the intended use case, this is considered a standard operational risk.

gcp-managed-airflow-recommendations