gcloud

Pass

Audited by Gen Agent Trust Hub on Mar 24, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches and runs the Google Cloud SDK installer from dl.google.com. Evidence: curl -O https://dl.google.com/dl/cloudsdk/channels/rapid/downloads/google-cloud-cli-linux-x86_64.tar.gz and ./google-cloud-sdk/install.sh in SKILL.md.
  • [EXTERNAL_DOWNLOADS]: Installs the firebase-tools package from the npm registry. Evidence: npm install -g firebase-tools in SKILL.md.
  • [COMMAND_EXECUTION]: Executes shell commands to manage infrastructure, including remote command execution on VM instances via SSH. Evidence: Multiple gcloud and firebase commands in SKILL.md.
  • [DATA_EXFILTRATION]: Provides instructions for accessing sensitive cloud information, such as secrets and system logs, and includes commands for modifying bucket permissions. Evidence: gcloud secrets versions access, gcloud logging read, and gsutil iam ch allUsers:objectViewer in SKILL.md.
  • [PROMPT_INJECTION]: Identifies an indirect prompt injection surface (Category 8) when processing external data sources. Evidence: 1. Ingestion points: gcloud logging read, gcloud compute instances get-serial-port-output, gcloud secrets versions access in SKILL.md. 2. Capability inventory: System command execution, cloud administrative access, remote SSH execution. 3. Boundary markers: Absent. 4. Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 24, 2026, 12:25 AM