huawei-cloud-optv-evolve-management
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes the Huawei Cloud KooCLI (
hcloud) and local helper scripts usingskill action=exec. This is the primary method for the agent to interact with the cloud service and perform local tasks like zipping code and managing local state caches. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates the evolution of algorithm code by sending user-provided scripts and natural language prompts to a cloud-based LLM. While the agent processes this untrusted content, the actual execution of the evolved code occurs within the sandboxed environment of the Huawei Cloud platform, not locally on the agent's host.
- [SAFE]: The skill adheres to security best practices by providing explicit warnings against credential exposure, including environment verification scripts (
precheck.shandprecheck.ps1), and utilizing a dedicated.stats/directory for local caching of project identifiers. All external references target official Huawei Cloud domains.
Audit Metadata