The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the use of the bl (Bailian CLI) to perform operations such as authentication, dataset validation, fine-tuning, and deployment. These are the primary functions of the skill and are performed through standard shell commands.
[COMMAND_EXECUTION]: Includes a localized monitoring utility, scripts/wait.sh, which is executed by the agent to poll for task completion. The script uses standard POSIX-compliant utilities (grep, cut, sleep) and is designed to handle shell environment limitations (such as zsh read-only variables) safely.
[EXTERNAL_DOWNLOADS]: Mentions the optional use of the edge-tts Python package for generating synthetic audio data for testing purposes. This is documented as a manual step for the user/agent to prepare data.
[DATA_EXFILTRATION]: Demonstrates legitimate interaction with the official dashscope.aliyuncs.com API for model inference. The interaction is secured using the DASHSCOPE_API_KEY environment variable, following industry-standard authentication practices.
[SAFE]: The skill implements proactive security and cost-management measures, including verifying authentication status before operations and requiring explicit human-in-the-loop confirmation before allocating reserved (billable) resources like mu or ptu plans.

bailian-train-deploy