The Agent Skills Directory

[COMMAND_EXECUTION]: The Python scripts scripts/agent.py and scripts/process.py utilize the subprocess module to programmatically interact with the gcloud CLI.
scripts/agent.py executes gcloud auth print-access-token to retrieve temporary authentication credentials for Google API requests.
scripts/process.py executes multiple gcloud commands to list backend services, retrieve IAM policies, enumerate access levels, and analyze device compliance.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by ingesting and processing metadata from external Google Cloud resources.
Ingestion points: Data is pulled from GCP via gcloud JSON outputs and HTTP responses from googleapis.com, including resource names, user emails, and audit log entries.
Boundary markers: The scripts do not implement explicit delimiters or warnings to ignore instructions embedded within the ingested metadata when generating reports.
Capability inventory: The tool chain includes local file-writing capabilities (beyondcorp_report.json, text-based audit reports) and the ability to execute further CLI queries based on project and policy identifiers.
Sanitization: No specific filtering or escaping is applied to resource identifiers or log content before inclusion in output files, though the risk is localized to the generated reports.

implementing-beyondcorp-zero-trust-access-model