nlm-skill

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly instructs extracting browser cookies and passing them verbatim (e.g., mcp__notebooklm-mcp__save_auth_tokens(cookies="<cookie_header>")), which requires the LLM to handle and output secret cookie/header values directly, creating an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow can ingest outsider-authored free text from public web pages during nlm source add --url ... and/or nlm research start ... (web research), which then becomes LLM-readable source content used in later nlm notebook query / generation calls.