The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run within its environment setup and execution scripts (run.py, __init__.py, setup_environment.py). These commands are used for virtual environment management, dependency installation, and executing internal Python logic.
[EXTERNAL_DOWNLOADS]: The skill automates the installation of required Python packages via pip and downloads the Google Chrome browser using the Patchright automation framework to ensure consistent fingerprinting and anti-detection capabilities.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting and processing content from the external NotebookLM service.
Ingestion points: Research answers derived from user-uploaded documents are retrieved in ask_question.py and browser_session.py.
Boundary markers: Output is returned as plain text without specialized delimiters or boundary instructions for the agent.
Capability inventory: The skill allows the execution of local scripts and the management of a local metadata library.
Sanitization: Content retrieved from the external service is not sanitized or filtered before being presented to the agent.

blog-notebooklm