marketing-pipeline-ai-content-automation

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

• Third-party content exposure detected (high risk: 0.95). The required runtime workflow calls autoResearch(...) to crawl external sources (e.g., TechCrunch/a16z/Twitter/LinkedIn) and then passes the resulting extracted text/insights as research.processedInsights into generateContent(...), which is embedded into the LLM prompt context (e.g., Research Context:\n${context} / JSON.stringify(research)), so outsider-authored web content can become LLM-readable context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's autoResearch and researchConfig explicitly fetch live content at runtime from external sites (e.g., https://techcrunch.com, https://a16z.com/blog, and Twitter/X/LinkedIn sources) and then injects that fetched content directly into AI prompts (generateWithClaude/generateWithOpenAI), so these URLs are runtime dependencies that control the agent's prompts.