hugging-face-model-trainer

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill repeatedly instructs the agent to include the user's HF_TOKEN in job submission payloads using the "$HF_TOKEN" syntax and explicitly states that "$HF_TOKEN" references your actual token value, which requires the LLM to handle or embed the secret verbatim in generated code/commands.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill includes examples that fetch and execute remote Python scripts at runtime (e.g., running hf_jobs with https://huggingface.co/datasets/mcp-tools/skills/raw/main/dataset_inspector.py and the GGUF conversion script that performs git clone https://github.com/ggerganov/llama.cpp.git), and it also references running TRL example scripts from URLs (e.g., https://raw.githubusercontent.com/huggingface/trl/main/examples/scripts/grpo.py and https://github.com/huggingface/trl/blob/main/trl/scripts/sft.py), which is clear evidence of runtime external code fetching and execution.