vertex-ai-endpoint-config

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [NO_CODE] (SAFE): The provided skill consists solely of a markdown configuration file (SKILL.md). There are no accompanying Python scripts, Node.js modules, or shell scripts.
  • [COMMAND_EXECUTION] (SAFE): The skill metadata requests access to Bash(gcloud:*). While this allows execution of Google Cloud commands, there is no logic within the skill that performs any unauthorized or dangerous command execution. The permission is appropriate for the stated purpose of managing Vertex AI endpoints.
  • [DATA_EXFILTRATION] (SAFE): No patterns for accessing sensitive local files (like SSH keys or AWS credentials) or exfiltrating data to external servers were found.
  • [REMOTE_CODE_EXECUTION] (SAFE): No external scripts are downloaded or executed. No package managers (pip, npm) are invoked.
  • [PROMPT_INJECTION] (SAFE): The instructional content is standard and does not contain attempts to bypass AI safety filters or override system-level instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:13 PM