The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references container images from NVIDIA's official NGC registry (nvcr.io) and Python wheels from established repositories like download.pytorch.org and wheels.vllm.ai. These are well-known and trusted infrastructure providers for the machine learning community.
[COMMAND_EXECUTION]: Instructions include standard shell commands for pip and docker. The pip commands utilize explicit version pinning and the --no-deps flag to prevent the resolution of incompatible or untrusted dependency trees. The docker commands use standard flags for GPU acceleration and memory management appropriate for the described hardware.
[DATA_EXFILTRATION]: No patterns of sensitive data access or exfiltration were detected. Network activities are restricted to software acquisition from established registries.
[PROMPT_INJECTION]: The skill contains technical documentation and instructions without any attempts to override model behavior, extract system prompts, or bypass safety filters.
[SAFE]: The skill follows security best practices for environment setup, such as recommending the use of pinned digests for Docker images to ensure reproducibility and supply chain integrity.

spark-environment-setup