Notebooklm mcp cli

Skill onfire7777/universal-ai-skills-library/skills/notebooklm-mcp-cli

Router-first AI skill system for Codex, Claude, Cursor, Hermes, Paperclip, OpenCode, and local AI stacks: search, preflight-route, and load 1,812 skills on demand without duplicating the corpus.

Install

npx -y skills add onfire7777/universal-ai-skills-library --skill notebooklm-mcp-cli

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

13 stars13 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when working with Google NotebookLM through jacob-bd/notebooklm-mcp-cli: listing or creating notebooks, adding sources, querying notebooks, generating audio or studio artifacts, revising slides, downloading artifacts, sharing notebooks, syncing Drive sources, running cross-notebook queries, configuring the NotebookLM MCP server, or diagnosing NotebookLM CLI authentication. Prefer this over generic Google or MCP skills when the task is specifically about NotebookLM.

SKILL.md

4.3 KB, 852 tokens by cl100k_base, as published. Nobody here has run it

NotebookLM MCP CLI

Source

Upstream repository: https://github.com/jacob-bd/notebooklm-mcp-cli
Local checkout: %USERPROFILE%\.notebooklm-mcp-cli\notebooklm-mcp-cli
Installed package: notebooklm-mcp-cli
CLI command: nlm
MCP server command: notebooklm-mcp
Universal stack source registry: %USERPROFILE%\.universal-ai-stack\config\source-integrations.json

Operating Model

Use the CLI first. The MCP server is available for clients that require a live tool endpoint, but it should not be started or registered by default for simple one-shot NotebookLM work.

Do not copy this upstream repository, cookies, browser profiles, generated media, or NotebookLM session state into AI roots or into this repo. The universal install uses:

the PyPI/uv tool install for executables,
one external source checkout for README-grounded reference,
this one canonical skill for routing and workflow instructions,
compact skill-router wrappers in each AI client.

Authentication

NotebookLM access depends on the user's Google account session.

Check auth before doing NotebookLM operations:

nlm login --check

If auth is missing, use the upstream login flow:

nlm login

For multiple Google accounts:

nlm login --profile work
nlm login --profile personal
nlm login profile list
nlm login switch work

Never print, paste, store, or commit browser cookies, Google OAuth state, downloaded private source files, or NotebookLM-generated private artifacts.

CLI Workflows

Use nlm --ai for the upstream's full AI-assistant command reference when exact syntax is needed.

Common commands:

nlm notebook list
nlm notebook create "Research Project"
nlm source add <notebook> --url "https://example.com/article"
nlm source add <notebook> --file "C:\path\source.pdf"
nlm notebook query <notebook> "What are the key findings?"
nlm cross query "Compare these notebooks on the main risks"
nlm studio create <notebook> audio --confirm
nlm download audio <notebook> <artifact-id>
nlm share public <notebook>
nlm doctor

Capabilities to prefer this skill for:

notebook list/create/delete workflows,
adding URL, text, Drive, and file sources,
source-grounded notebook queries,
cross-notebook research,
batch query/create/delete operations,
pipeline run/list workflows,
tags and smart notebook selection,
audio overview, video, slide, and other Studio artifacts,
artifact download,
public or invite sharing,
Google Drive source sync,
NotebookLM MCP configuration and troubleshooting.

MCP Configuration

Only configure MCP when the target AI client explicitly needs an MCP server. For universal stack work, prefer direct CLI calls or skill-router routing.

Generate a generic MCP config when needed:

nlm setup add json

Use upstream client-specific setup only after checking existing client config to avoid duplicate MCP entries:

nlm setup list
nlm setup add claude-code
nlm setup add gemini
nlm setup add cursor

If a client already has the compact universal skill-router adapter, do not install a full duplicate NotebookLM skill tree into that client. Keep the MCP server as an optional stdio command:

notebooklm-mcp

Quality Gates

Before reporting this integration healthy, verify:

nlm --version
notebooklm-mcp --help
nlm doctor
skill-router skill notebooklm-mcp-cli
skill-router preflight --json "query my NotebookLM notebook about this research"
skill-router skills validate-manifest

nlm doctor may report that Google login is missing. That is an auth state, not a failed installation, unless the user expected the account to be logged in already.