gcp-logs

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • COMMAND_EXECUTION (SAFE): The skill provides structured gcloud command templates for log retrieval. This is the primary intended purpose and does not involve shell injection or arbitrary command execution from untrusted inputs.
  • INDIRECT_PROMPT_INJECTION (LOW): The skill is designed to ingest external log data, which constitutes an untrusted data source.
  • Ingestion points: Logs ingested via gcloud logging read as specified in SKILL.md.
  • Boundary markers: Absent; log data is not wrapped in protective delimiters in the provided templates.
  • Capability inventory: The skill facilitates execution of gcloud and jq commands.
  • Sanitization: No sanitization or filtering of log content for instruction-like patterns is provided.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:32 PM