transformers-js

Pass

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides instructions for installing official Hugging Face packages (@huggingface/transformers, @huggingface/tokenizers) via npm and loading library assets from the JSDelivr CDN. These are established, well-known services.
  • [EXTERNAL_DOWNLOADS]: The skill's primary function involves downloading machine learning models (ONNX format) from the Hugging Face Hub (huggingface.co). These downloads are fundamental to the stated purpose of the skill and target a trusted organization.
  • [DATA_EXFILTRATION]: Documentation in references/CONFIGURATION.md and references/CACHE.md provides templates for managing authentication (e.g., HF_TOKEN) and custom caching (e.g., S3). These are documented as configuration options for the developer and do not contain hardcoded secrets or malicious exfiltration logic.
  • [COMMAND_EXECUTION]: The skill includes standard terminal commands for package installation and CLI usage examples. All commands are consistent with the development workflows of the Transformers.js library.
  • [PROMPT_INJECTION]: As a machine learning tool, the skill processes external input data (text, images, audio). While this represents a surface for indirect prompt injection, it is the intended functionality of the integration, and the documentation includes configuration options to mitigate risks (such as local_files_only mode).
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 14, 2026, 10:58 AM
Security Audit — agent-trust-hub — transformers-js