compute-future-reservations
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- Command Execution: The skill involves the use of gcloud compute commands to interact with cloud infrastructure. These tools are used to assess feasibility and create reservations, and the skill correctly emphasizes the need for human approval before performing mutating actions.
- Indirect Prompt Injection: The instructions incorporate user-defined inputs, such as project identifiers and resource specifications, into shell command templates. This presents a surface for potential input manipulation. The skill mitigates this by instructing the agent to verify parameters and request explicit user confirmation.
- External Resource References: The skill links to official Google Cloud documentation for hardware limits and pricing. These references are to a well-known service and provide necessary context for the user without introducing external risk.
Audit Metadata