skills/melvynx/cli-skills/gcloud/Gen Agent Trust Hub

gcloud

Pass

Audited by Gen Agent Trust Hub on Nov 21, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs the official Google Cloud SDK using Homebrew. This is a reference to a well-known service and a standard installation path for developer tools.
  • [COMMAND_EXECUTION]: The skill is primarily a wrapper for shell-based gcloud CLI commands. These commands are used for infrastructure management (Compute Engine, Cloud Run, Storage, etc.) and are used as intended for this tool's category.
  • [INDIRECT_PROMPT_INJECTION]: The skill includes commands to read external data which could potentially contain malicious instructions intended for the agent.
  • Ingestion points: gcloud functions logs read <name> and gcloud app logs tail in SKILL.md allow the agent to ingest log data from external cloud environments.
  • Boundary markers: No explicit delimiters or instructions to ignore embedded content are provided when reading logs.
  • Capability inventory: The skill possesses extensive capabilities including shell command execution (gcloud), file system operations (gcloud storage cp), and cloud resource modification (gcloud compute instances delete).
  • Sanitization: No sanitization or validation of the log content is performed before it enters the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 21, 2026, 07:45 AM