The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is primarily designed to guide an agent in executing a wide range of shell commands via the hf CLI, covering authentication, repository management, and compute tasks.
[EXTERNAL_DOWNLOADS]: Facilitates the downloading of models, datasets, and container images from Hugging Face Hub (huggingface.co), which is a well-known service for AI assets.
[REMOTE_CODE_EXECUTION]: Provides instructions for running arbitrary code and scripts on Hugging Face's cloud infrastructure using hf jobs run and hf jobs uv run. This includes the capability to execute Python scripts directly from remote URLs (e.g., from huggingface.co or other locations).
[CREDENTIALS_UNSAFE]: Documents authentication workflows that involve Hugging Face tokens (HF_TOKEN). It provides examples of non-interactive login via CLI flags and passing secrets to cloud-hosted jobs, which are standard operations for this CLI tool.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes external content.
Ingestion points: Untrusted data enters the context through user-specified repository IDs in hf download and remote script URLs in hf jobs uv run.
Boundary markers: No specific boundary markers or instructions to disregard embedded commands in downloaded files are present.
Capability inventory: The CLI tool has the ability to read/write local files and execute code on remote infrastructure.
Sanitization: No explicit sanitization or validation of repository names or URLs is described.

hugging-face-cli