torchforge-rl-training

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

• Third-party content exposure detected (high risk: 0.80). The skill's required workflows explicitly ingest public datasets and models from HuggingFace (e.g., config/grpo_math.yaml: dataset.path: "openai/gsm8k", streaming: true and "Model from HuggingFace Hub" in prerequisites), so it reads and uses untrusted third-party content as part of runtime training that can materially influence model behavior.