search-first

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

• Third-party content exposure detected (high risk: 0.78). The required workflow explicitly includes “Parallel Search” across public sources (npm/PyPI/GitHub/Web) and “Install package / Configure MCP,” which at runtime can ingest outsider-authored free text such as package README/docs, search results, and scraped web content into the agent’s LLM context for evaluation/decision-making.