The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes subprocess.run in scripts/run.py, scripts/setup_environment.py, and scripts/__init__.py. These calls are dedicated to project orchestration, including the creation of virtual environments, dependency management with pip, and the execution of the skill's own internal script modules.
[EXTERNAL_DOWNLOADS]: During its setup phase, the skill fetches required Python packages from standard registries and triggers the download of browser binaries (Google Chrome or Chromium) via the patchright library. These operations target well-known services and are necessary for the skill's automation functionality.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from Google NotebookLM, creating an attack surface for indirect prompt injection if notebooks contain malicious instructions.
Ingestion points: External content is read through browser automation in scripts/ask_question.py.
Boundary markers: The SKILL.md frontmatter and mindset instructions, along with the FOLLOW_UP_REMINDER in scripts/ask_question.py, provide guidelines for the agent to analyze results and identify gaps.
Capability inventory: The skill has capabilities to write to local data/ and logs/ directories and execute project scripts via a wrapper.
Sanitization: The skill relies on the agent's synthesis and follow-up procedures to verify content before presenting it to the user.
[DYNAMIC_EXECUTION]: The skill employs a run.py wrapper to dynamically invoke its internal Python scripts. This orchestration involves constructing and executing command-line strings to run scripts within the project's local virtual environment based on user or agent input.

notebooklm