huawei-cloud-alarm-bind-inspection
Warn
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [DYNAMIC_EXECUTION]: The script 'scripts/patch_excel.py' is designed to programmatically read, modify, and overwrite 'scripts/generate_excel_report.py'. This self-modifying code pattern, while used here for merging reporting logic, constitutes dynamic execution and modification of the skill's own codebase.
- [COMMAND_EXECUTION]: Scripts such as 'scripts/config.py', 'scripts/enrich_metric_dimension.py', and 'scripts/main_aom_alarm.py' execute shell commands via 'subprocess.run' to interact with the 'hcloud' CLI tool. While these are legitimate calls to official vendor utilities, the pattern enables arbitrary command execution.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Excel assets ('assets/关键资源标记表.xlsx' and 'assets/华为云租户面监控告警基线-V26.01-beta.xlsx'). The lack of sanitization or boundary markers when processing these cell values creates a surface where malicious content in the spreadsheets could influence the agent's logic or output.
- [EXTERNAL_DOWNLOADS]: The 'references/cli-installation-guide.md' file references the download and execution of a bash script from a Huawei Cloud OBS bucket ('myhuaweicloud.com'). As this domain belongs to the official vendor, the reference is documented neutrally, though the use of piped execution ('curl | bash') is a high-risk installation method.
Audit Metadata