ai-infrastructure-huggingface-inference

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill documents patterns for processing external data (text, images, audio) through remote AI models via the Hugging Face Inference SDK, which naturally creates a surface for indirect prompt injection.
  • Ingestion points: Code examples in examples/core.md (e.g., chatCompletion) and examples/tasks.md (e.g., textToImage, automaticSpeechRecognition) demonstrate methods for processing external inputs.
  • Boundary markers: The instructions do not define specific delimiters or boundary instructions to isolate user-supplied data from system prompts, although this is typical for low-level SDK patterns.
  • Capability inventory: The skill enables network communication with the Hugging Face Hub and various third-party inference providers (such as Groq and Cerebras) through the @huggingface/inference library.
  • Sanitization: The patterns focus on API implementation rather than input validation or sanitization logic.
  • [EXTERNAL_DOWNLOADS]: The skill references the installation of an external library essential for its functionality.
  • Evidence: reference.md specifies npm install @huggingface/inference. This is the official, well-known SDK provided by Hugging Face, a recognized and trusted service.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 03:56 PM
Security Audit — agent-trust-hub — ai-infrastructure-huggingface-inference