NotebookLM Skill for Claude Code

August 6, 2026 · View on GitHub

NotebookLM Skill for Claude Code

A Claude Code / Agent Skill that lets your AI agent query your Google NotebookLM notebooks — citation-backed, source-grounded answers from your own documents, plus notebook / source / Studio management. Zero hallucinations: answers come only from your sources.


This skill drives the @roomi-fields/notebooklm-mcp engine — the fastest, most complete NotebookLM automation (dual internal-RPC + browser transport, 33 REST endpoints, full Studio generation, multi-account). It works two ways, whichever your session already has:

  • notebooklm MCP tools if configured, or
  • the engine's HTTP REST API via the bundled scripts/nblm.sh.

What it enables

  • Citation-backed Q&A — ask a notebook a question, get an answer grounded in your sources with the exact cited passages.
  • Studio generation — audio overview, report, video, infographic, presentation, data table, flashcards, quiz, mind map.
  • Source discovery — web/Drive research to find and import new sources.
  • Research at scale — the daily-quota-aware batch-to-cache pattern (ingest once with NotebookLM, then retrieve offline, unlimited).

Install

As a plugin (recommended — gets the skill and the MCP server):

/plugin marketplace add roomi-fields/claude-plugins
/plugin install notebooklm@roomi-fields

Or just the skill — copy skills/notebooklm/ into your skills folder:

git clone https://github.com/roomi-fields/notebooklm-skill.git
cp -r notebooklm-skill/skills/notebooklm ~/.claude/skills/

Then log in to Google once (a visible Chrome window opens):

notebooklm-mcp-setup-auth

Use

Just ask your agent things like:

"Ask my Thesis notebook what the main critiques of the framework are, with citations."

The skill handles auth checks, transport selection, citation formatting, and the quota-aware batch pattern. See skills/notebooklm/SKILL.md.

License

MIT — see LICENSE. Built by @roomi-fields.