The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill manages Google authentication using browser cookies and named profiles. It includes commands like nlm login for automated extraction and manual import from cookies.txt. This is a standard and necessary mechanism for the tool to interact with the NotebookLM service.
[DATA_EXFILTRATION]: The skill is designed to interact with external networks to manage notebooks and fetch research sources from URLs and Google Drive. It also accesses local files via the --file argument to upload them as sources. These behaviors are core to the skill's primary function.
[COMMAND_EXECUTION]: The skill documentation includes examples of integrating the nlm CLI tool using shell scripts and Python's subprocess module. These are standard implementation patterns for the provided utility.
[PROMPT_INJECTION]: The skill ingests untrusted external data from URLs and documents to provide research capabilities, which creates a surface for indirect prompt injection. This is a known risk for agents processing external content and is handled here as a functional requirement for research and summarization tasks.

notebooklm-mcp-programmatic-access