The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches AI model assets and configuration files from ModelScope, which is a well-known and reputable service for machine learning model hosting.
[COMMAND_EXECUTION]: Employs docker exec and docker cp to interact with containers, allowing the skill to deploy test scripts and execute inference commands in an isolated environment.
[REMOTE_CODE_EXECUTION]: Invokes vLLM with the --trust-remote-code parameter, a standard but high-privilege configuration that enables the execution of custom code provided within a model's repository; this risk is mitigated by targeting a well-known service (ModelScope).
[DYNAMIC_EXECUTION]: Dynamically generates and runs Python commands and shell instructions within the container based on user-supplied model identifiers and hardware settings.

model-verify-flagos