arkcli-infer-endpoint

Pass

Audited by Gen Agent Trust Hub on Jul 1, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill is designed to execute the arkcli binary to perform lifecycle management of inference endpoints, including operations such as creation, listing, retrieval of details, and controlling status (start/stop).
  • [DATA_EXFILTRATION]: While the skill manages cloud resources and interacts with API keys via CLI flags, it follows standard practices for cloud resource management. There are no patterns suggesting unauthorized data exfiltration or hardcoded credentials; instructions explicitly guide the use of SSO login or profile-based authentication.
  • [PROMPT_INJECTION]: The skill provides instructional guidance for interpreting user intents (e.g., mapping "create" requests to specific workflows) and includes safety constraints for voice-related models. These instructions align with the functional purpose of the skill and do not attempt to bypass agent safety filters or override core system prompts.
  • [EXTERNAL_DOWNLOADS]: The skill requires the arkcli tool but does not perform any runtime downloads of untrusted scripts or third-party packages. All dependencies and references are directed toward the vendor's ecosystem.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 1, 2026, 08:29 AM
Security Audit — agent-trust-hub — arkcli-infer-endpoint