cloud-scheduler-job-creator

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION] (HIGH): The skill is vulnerable to indirect prompt injection as it ingests untrusted user requests to perform actions with high-privilege tools.
  • Ingestion points: User requests triggered by phrases like "Help me with cloud scheduler job creator" (SKILL.md).
  • Boundary markers: Absent. There are no instructions to the agent to ignore embedded commands or treat user input as data only.
  • Capability inventory: Access to Bash(gcloud:*), Write, and Edit tools (SKILL.md).
  • Sanitization: Absent. No validation or filtering logic is provided to ensure generated gcloud commands are restricted to job creation.
  • [COMMAND_EXECUTION] (MEDIUM): The skill requests overly broad permissions using a wildcard in the bash tool.
  • Evidence: allowed-tools: Bash(gcloud:*) in SKILL.md. This allows the agent to execute any gcloud command, including deleting projects, modifying IAM policies, or exfiltrating data, which exceeds the least-privilege requirements for simply creating scheduler jobs.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 06:34 AM