yao-deepseek-crawler

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

• Third-party content exposure detected (high risk: 0.75). The workflow runs scripts/deepseek_batch_crawl.mjs to collect DeepSeek assistant DOM text and citation panels (outsider-authored web content) and then scripts/analyze_deepseek_results.py consumes that extracted prose for semantic review, so attacker-controlled free text from visited DeepSeek pages can flow into the analysis/LLM context.