The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install an external dependency, the 'NotebookLM MCP server', using a CLI command nlm setup add claude-code. This involves downloading and configuring third-party software not included in the skill itself.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data from various sources, creating a surface for indirect injection attacks.
Ingestion points: The skill uses source_add to import user-provided URLs and documents, and performs WebSearch calls to gather data from the public internet.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when processing the fetched content, increasing the risk that malicious instructions in source documents could be obeyed by the agent.
Capability inventory: The skill possesses the ability to write to the local file system (creating a research/ directory and markdown briefs) and perform network operations via the NotebookLM MCP tools.
Sanitization: There are no explicit sanitization or filtering steps mentioned for the external content before it is processed by the research engine or used to generate final reports.
[COMMAND_EXECUTION]: The skill relies on execution of several MCP tools (notebook_create, research_start, studio_create) to perform its core functions. While these are scoped to the research task, they represent a significant capability tier when combined with external data ingestion.

deep-research-notebooklm