always-verify-gcp
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill manages GCP resources using CLI tools via
run_shell_command. This core functionality is secured by a rigorous verification process requiring agent-led documentation checks and dry-run validation. - [INDIRECT_PROMPT_INJECTION]: The skill processes user requests for GCP actions. (1) Ingestion point: User input processed by the workflow. (2) Boundary markers: None, but the multi-step verification process acts as a functional gate. (3) Capability inventory:
run_shell_commandandsearch_documents. (4) Sanitization: Mandatory syntax verification against official documentation and a required dry-run step to preview impact. - [SAFE]: The skill implements significant safety guardrails, including tool-specific heuristics and structured error diagnostics. No obfuscation, persistence, or data exfiltration vectors were identified.
Audit Metadata