The Agent Skills Directory

[COMMAND_EXECUTION]: The script flexus_lifecycle.py utilizes subprocess.run to execute a local companion script, flexus_specs_extractor.py. This is an internal execution pattern used to fetch dynamic server specifications and does not involve running arbitrary shell commands from user input.
[EXTERNAL_DOWNLOADS]: The skill fetches HTML content from the official Huawei Cloud documentation at support.huaweicloud.com. This is a trusted vendor resource used to ensure that server specifications and pricing are up to date. While it involves external network requests, the source is authoritative and relevant to the skill's primary purpose.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from public web pages which constitutes a potential attack surface for indirect prompt injection.
Ingestion points: flexus_specs_extractor.py scrapes content from the Huawei Cloud support site using urllib.request.urlopen.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content embedded in the scraped HTML.
Capability inventory: The skill possesses high-privilege capabilities including instance creation, renewal, and cancellation via the hcss and bss APIs in flexus_lifecycle.py.
Sanitization: Content is parsed using regular expressions and converted into structured JSON before use, but there is no specific filtering for natural language instructions that might influence the agent's behavior.
[SAFE]: The skill includes robust security guidelines in SKILL.md, mandating that sensitive credentials (AK/SK) be passed via environment variables and strictly forbidding the agent from printing these values in the conversation or logs. Additionally, the skill implements a mandatory human-in-the-loop confirmation mechanism for all operations involving financial costs.

huawei-cloud-flexus-l-server-manage