huawei-cloud-agentarts-obs-diagnosis
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is built around the execution of shell commands using the official Huawei Cloud CLI (
hcloud). These commands are used to query metadata and data from services including APM (Application Performance Management), AOM (Application Operations Management), and LTS (Log Tank Service). - [EXTERNAL_DOWNLOADS]: The skill references the official Huawei Cloud support documentation for the installation of the KooCLI tool. These references point to legitimate vendor-controlled domains (
support.huaweicloud.com). - [INDIRECT_PROMPT_INJECTION]: The skill processes external data retrieved from cloud services, which presents a surface for indirect prompt injection.
- Ingestion points: Data ingested from logs via
hcloud LTS ListLogs, metrics viahcloud AOM ListInstantQueryAomPromGet, and traces viahcloud APM SearchTransaction. - Boundary markers: No explicit markers are used to delimit external data within the diagnostic workflow.
- Capability inventory: The skill can execute CLI commands and perform network operations via
curlto theAGENT_OPS_ENDPOINT. - Sanitization: The skill identifies common error codes (e.g.,
LTS.2446) but does not explicitly detail sanitization of raw log or metric content before inclusion in diagnostic reports.
Audit Metadata