agent-platform-access-audit
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- Standard Command Execution: The skill utilizes the
gcloudandbqcommand-line tools to interact with Google Cloud services. This is consistent with its intended purpose of infrastructure auditing and data retrieval from BigQuery. - SQL Injection Prevention: The instructions include a security-focused workflow for constructing SQL queries. It mandates the use of named query parameters (
--parameter) and regex validation (^[a-z0-9_-]+$) for project and dataset identifiers to prevent SQL injection vulnerabilities when dealing with user-supplied or discovered values. - Untrusted Data Handling: The skill acknowledges that log data (such as resource names or error messages) can be controlled by external actors. It provides explicit 'Untrusted Log Data Guardrails' instructing the agent to treat these values strictly as data and never as instructions, which helps mitigate indirect prompt injection risks.
- Least Privilege and Guidance: The workflow directs the agent to inform users about missing permissions or APIs rather than attempting to self-remediate, maintaining a clear boundary of authorization and adhering to the principle of least privilege.
Audit Metadata