notebooklm-research

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). Multiple URLs point to an unvetted third‑party GitHub user/repository (release tags, clone URLs, compare links and instructions to run install.sh/./install) which is a common vector for distributing arbitrary scripts or installers — review code and release artifacts before running or installing.

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

• Third-party content exposure detected (high risk: 0.85). Yes: the required runtime workflow performs web/RSS/URL and “trend-pulse” ingestion, which brings back outsider-authored page text/research reports that are then used as contextual input for subsequent LLM-driven “ask”/content generation in the pipeline (SKILL.md describes URL/RSS/trend-to-content ingestion into NotebookLM, whose outputs are then handed to Claude).