rag-pipeline

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

• Third-party content exposure detected (high risk: 0.95). Bright Data Discover is used at runtime with includeContent: true, and it fetches/parses page content from public web sources (outsider-authored text) which is then inserted into the LLM prompt context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill calls Bright Data's Discover endpoint (e.g., https://api.brightdata.com/discover) at runtime to fetch web page content which is then injected into assembled prompts (buildPrompt/retrieve), so externally-fetched content directly controls the agent's prompt context and is a required runtime dependency.