The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes bl, npm, and gh commands to initialize skills, manage configurations, and facilitate issue reporting to GitHub.
[EXTERNAL_DOWNLOADS]: Fetches installation scripts and CLI updates from official Alibaba Cloud domains and the npm registry.
[REMOTE_CODE_EXECUTION]: Executes remote installation scripts from Aliyun via curl | bash or Invoke-Expression (iex). These are the documented installation paths for the official CLI.
[CREDENTIALS_UNSAFE]: Instructs the agent on how to manage API keys and Alibaba Cloud AccessKeys. It provides explicit guidelines and mandatory redaction rules to prevent the leakage of these credentials during troubleshooting or bug reporting.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied messages and prompts by passing them to the bl CLI for processing by AI models.
Ingestion points: User input is interpolated into flags like --message and --prompt in various bl commands (e.g., bl text chat, bl image generate).
Boundary markers: Utilizes standard command-line flags to separate user-provided content from instructions.
Capability inventory: The CLI has broad capabilities including network access, file system interaction (writing generated media), and cloud resource management.
Sanitization: Implements strict redaction protocols for secrets and account identifiers before data is shared outside the local environment.

bailian-protocol