The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the official Alibaba Cloud CLI (aliyun) via subprocess.run to verify identity (sts get-caller-identity) and decode diagnostic messages (ram decode-diagnostic-message). The commands are constructed using list-based arguments, which prevents shell injection. This is a standard and safe way for a skill to interact with local system tools.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface by processing external, untrusted data such as OSS Bucket Policy documents and EncodedDiagnosticMessage strings from error responses. However, the skill is designed to be strictly read-only, only outputting findings and templates for manual user application, which mitigates the risk of this surface being used to drive unintended mutating actions.
Ingestion points: _oss_client.py (via GetBucketPolicy) and oss_cross_account_auth_diagnosis.py (via --encoded-diagnostic-message argument).
Boundary markers: The skill does not use explicit boundary markers in the processed data, but it uses structured JSON parsing to separate data from logic.
Capability inventory: The skill is restricted to read-only metadata queries via the oss2 SDK and identity/decoding checks via the aliyun CLI. It has no write or delete capabilities.
Sanitization: The skill uses standard JSON deserialization to process policy documents and error messages.
[EXTERNAL_DOWNLOADS]: The _doc_lookup.py script fetches documentation excerpts from the official Alibaba Cloud Help Center (help.aliyun.com). These network operations are limited to the vendor's own official domains and are used to provide the user with verified reference material, qualifying as a safe and expected behavior.

alibabacloud-oss-cross-account-auth-diagnosis