huggingface-accelerate

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

• Third-party content exposure detected (high risk: 0.70). The skill's examples explicitly load public, user-generated data (e.g., references/performance.md shows load_dataset("openwebtext")) and instruct cloning public repos (references/megatron-integration.md git clone), so the training workflows can ingest untrusted third‑party content as part of runtime processing.