transformers-js

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

• Third-party content exposure detected (medium risk: 0.30). The runtime ingests outsider-provided free text as the model input to pipeline(...)(text)/generator(prompt) in the required workflow examples (e.g., server req.body.text), so an outsider can supply arbitrary text that the LLM processes.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill contains browser/module imports that fetch and execute remote code at runtime from CDNs (e.g., https://cdn.jsdelivr.net/npm/@huggingface/transformers, https://cdn.jsdelivr.net/npm/@huggingface/transformers@4, and https://cdn.jsdelivr.net/npm/onnxruntime-web/dist/), which are runtime external dependencies that will execute remote JavaScript/WASM.