The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/e2e_test.py utilizes subprocess.run to execute the skill's main entry script scripts/flink_ververica_ops.py for automated testing purposes. This execution is limited to internal code and is part of the development verification suite.
[EXTERNAL_DOWNLOADS]: The references/cli-installation-guide.md file includes instructions to install the Alibaba Cloud CLI via curl -fsSL https://aliyuncli.alicdn.com/install.sh | bash. As the source is an official vendor domain and the author is identified as the vendor, this is a trusted installation method.
[PERSISTENCE]: The references/cli-installation-guide.md file guides users to persist their environment configuration by adding Alibaba Cloud credentials to shell profiles such as ~/.bashrc or ~/.zshrc. This is standard practice for configuring CLI tools and is explicitly recognized as safe secret management in this context.
[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by ingesting and processing user-supplied SQL and JSON data through cloud APIs. However, it mitigates these risks by implementing mandatory confirmation gates (--confirm) for all mutating operations.
Ingestion points: User-provided SQL statements and JSON request bodies; local input files for batch validation.
Boundary markers: Explicit safety protocols and mandatory --confirm flag requirements for destructive actions.
Capability inventory: Execution of cloud API mutations and local file I/O via the provided toolset.
Sanitization: Relies on the structure of the Alibaba Cloud SDK and JSON parsing, though it does not perform deep content validation of raw SQL strings at the script level.

alibabacloud-flink-workspace-ops