The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of the Aliyun Bailian (bl) CLI tool. This includes commands for authentication (bl auth login), configuration (bl config set), and functional tasks like text generation or image analysis. These operations are the primary intended purpose of the skill family.
[EXTERNAL_DOWNLOADS]: The skill provides instructions for installing and updating the CLI tool using standard package managers (npm install -g bailian-cli) and official skill repositories (npx skills add modelstudioai/cli). Updates via bl update require explicit user confirmation.
[DATA_EXFILTRATION]: The skill includes a detailed protocol for reporting bugs to a GitHub repository (modelstudioai/cli). Crucially, it mandates an 'Agent Pre-flight Checklist' for redaction, explicitly instructing the agent to remove API keys, tokens, account IDs, and business-sensitive data from issue templates before submission. It also requires the agent to show the redacted content to the user for final approval before submission.
[CREDENTIALS_UNSAFE]: The skill provides instructions for managing authentication credentials (API keys). It correctly advises using environment variables (DASHSCOPE_API_KEY) or the tool's built-in login mechanism, rather than hardcoding secrets. It also includes instructions for clearing credentials (bl auth logout).
[INDIRECT_PROMPT_INJECTION]: While the skill processes user prompts for transmission to the Bailian API, it implements safety measures such as explicit system prompts to control output behavior (e.g., forcing language selection) and clear boundaries between user content and CLI parameters.

bailian-protocol