hf-cli

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The skill is the Hugging Face CLI wrapper; at runtime, commands like hf download REPO_ID / hf datasets info / hf spaces info fetch outsider-authored public Hub content (repo/dataset/space files and metadata) over the network and then ingest it as readable text/JSON into the agent’s execution context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill includes installation commands that download and immediately execute remote scripts (curl -LsSf https://hf.co/cli/install.sh | bash -s and curl -fsSL https://raw.githubusercontent.com/huggingface/hf-mount/main/install.sh | sh), which fetch and run remote code as required setup steps.