The Agent Skills Directory

[SAFE]: The skill serves as a technical reference guide for Ray LLM 2.57.0, providing correct idioms for deployment and batch inference. All referenced URLs point to the official ray-project GitHub repository and official Ray documentation.
[EXTERNAL_DOWNLOADS]: The skill recommends installing the ray[llm] extra via pip. This is the standard, official method for installing Ray with its matched vLLM dependency from the Python Package Index (PyPI).
[COMMAND_EXECUTION]: The provided code examples demonstrate the use of Ray's declarative Python APIs for model serving and batch processing. These examples do not include any arbitrary command execution, shell spawning, or system-level modifications.
[PROMPT_INJECTION]: The skill includes code templates for batch inference and serving that involve string interpolation of data into prompts (e.g., for summarization). These are standard educational examples for using LLM libraries and do not contain instructions designed to bypass safety filters or override agent behavior. The skill correctly notes that these configs can be validated on CPU-only hardware.

ray-llm