firecrawl-scrape

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

• Third-party content exposure detected (high risk: 0.75). 该技能的运行时会接收用户提供的 materialUrl/materialText/materialFile，其中 materialUrl 对应“运行时抓取的公开网页正文”（外部网页内容属于 outsider），并将抓取到的可读文本注入到技能的上下文用于 LLM 处理。

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

• Potentially malicious external URL detected (high risk: 1.00). The runner calls the remote AI Skills API at https://ai-skills.ai/api/execute (DEFAULT_BASE_URL) at runtime to execute the skill, so this external endpoint is a required runtime dependency that performs remote execution/controls the skill's behavior.