dspy-vllm

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

• Third-party content exposure detected (high risk: 0.80). SKILL.md explicitly notes that "vLLM downloads the model from HuggingFace on first start" (Gotchas section), meaning the skill will fetch user-hosted public model repositories from HuggingFace — untrusted third-party content — which the agent will load and whose behavior can materially influence subsequent decisions and tool use.