The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the Alibaba Cloud CLI tool (aliyun) using subprocess.run in scripts/_oss_client.py for identity verification. The implementation uses a fixed argument list and redirects stdin to subprocess.DEVNULL, which is a secure method for calling external binaries.\n- [COMMAND_EXECUTION]: The SKILL.md instructions guide the agent to verify and potentially upgrade the aliyun CLI using the upgrade --yes command. This is a standard management action for vendor-provided tooling and ensures compatibility for identity checks.\n- [EXTERNAL_DOWNLOADS]: The skill downloads documentation metadata and content from the official Alibaba Cloud help center (help.aliyun.com) through scripts/_doc_lookup.py. These operations are limited to the vendor's own official infrastructure and are used to provide verified guidance.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user-supplied symptoms and questions, as well as external documentation content, constituting a vulnerability surface that is well-mitigated by strict operational constraints.\n
Ingestion points: User inputs for the --symptom and --question parameters and external markdown content retrieved from help.aliyun.com.\n
Boundary markers: The skill defines "Absolute Rules" in SKILL.md to prohibit all non-read-only actions and uses structured JSON output to ensure deterministic behavior.\n
Capability inventory: The skill can execute the aliyun CLI for identity verification and issue read-only metadata queries (e.g., GetBucketInfo, ListBuckets) to the OSS control plane.\n
Sanitization: External documentation content is programmatically cleaned of markdown syntax and truncated in scripts/_doc_lookup.py before being used.

alibabacloud-oss-ossfs-mount-diagnosis