notemdpro-system-architecture

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

• Third-party content exposure detected (high risk: 0.90). The skill's _performResearch workflow (used by generateContentForTitle) calls SearchManager → TavilyProvider/DuckDuckGoProvider and uses fetchContentFromUrl to ingest and combine public web page content into the LLM context, so arbitrary third‑party webpages can directly influence subsequent tool calls and decisions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's _performResearch flow (SearchManager → DuckDuckGoProvider → fetchContentFromUrl) fetches arbitrary external URLs (DuckDuckGo search result pages) at runtime and injects their content into the combined research context used as LLM prompts, which directly controls model input.