NotebookLM SkillSAFE
NotebookLM does the research, Claude writes the content. Research → Synthesis → Content Creation → Publishing. Claude Code Skill + MCP Server.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
Source-grounded NotebookLM automation for terminals and AI agents.
[](https://github.com/claude-world/notebooklm-skill/actions/workflows/ci.yml) [](https://pypi.org/project/notebooklm-skill/) [](https://pypi.org/project/notebooklm-skill/) [](LICENSE)
繁體中文
notebooklm-skill gives humans and MCP clients one consistent interface for Google NotebookLM. It creates notebooks from URLs, raw text, and local files; asks grounded questions with citation metadata; completes fast or deep web research; and generates or downloads NotebookLM artifacts.
The project is built around notebooklm-py 0.7.x and includes:
- a JSON-first core CLI;
- five end-to-end research pipelines;
- a 13-tool FastMCP server;
- profile-aware authentication and Skill installers;
- one shared compatibility layer, so CLI, pipelines, and MCP use the same behavior.
This is an unofficial integration with NotebookLM's web API. Google can change the service, availability, quotas, or artifact behavior without notice.
Quick start
Isolated source install
The installer creates a dedicated virtual environment, installs Chromium, links five commands into ~/.local/bin, and installs the Claude Code Skill using the standard directory layout.
git clone https://github.com/claude-world/notebooklm-skill.git cd notebooklm-skill ./install.sh notebooklm-auth setup notebooklm-skill list
Ensure ~/.local/bin is on PATH.
PyPI or uvx
# Persistent virtual environment python3 -m venv .venv source .venv/bin/activate python -m pip install notebooklm-skill python -m playwright install chromium notebooklm-auth setup # Or run without a persistent install uvx --from no
addac501085eOBSERVED · 2026-10-01Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add notebooklm-skill --env NOTEBOOKLM_AUTH_JSON=${NOTEBOOKLM_AUTH_JSON} -- uvx notebooklm-skill{
"mcpServers": {
"notebooklm-skill": {
"command": "uvx",
"args": [
"notebooklm-skill"
],
"env": {
"NOTEBOOKLM_AUTH_JSON": "${NOTEBOOKLM_AUTH_JSON}"
}
}
}
}Exposed tools (13)
7 read · 5 write · 1 destructive. Blast radius: 1 tool can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
nlm_add_source | write | Add exactly one URL, text, or local file source to a notebook. |
nlm_ask | read | Ask a source-grounded question and return answer citations. |
nlm_create_notebook | write | Create a notebook with optional URL, raw-text, and local-file sources. |
nlm_delete | destructive | Permanently delete a notebook only when ``confirm`` is true. |
nlm_download | read | Download an exact artifact ID (or latest) to a local path. |
nlm_generate | read | Generate an artifact, optionally waiting for completion. |
nlm_list | read | List NotebookLM notebooks with IDs, titles, and source counts. |
nlm_list_artifacts | read | List actual generated artifacts, optionally filtered by canonical type. |
nlm_list_sources | read | List ingested sources with current kind and processing status. |
nlm_research | write | Run pinned web research, wait for a terminal state, and import findings. |
nlm_research_pipeline | write | Create a notebook, ingest URLs, ask questions, and assemble content. |
nlm_summarize | read | Return the NotebookLM-generated summary for a notebook. |
nlm_trend_research | write | Fetch trend-pulse topics, research/import each, and create grounded drafts. |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (1 observation(s))
- Network
- declared (1 observation(s))
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (2)
nlm_delete
notebooklm-py, fastmcp, python-dotenv, httpx, feedparser
Gates applied: no_behavioural_pass.
addac501085efull audit observations/trust-audit/mcp-server/claude-world__notebooklm-skill.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-01 | addac501085e | SAFE | B | 89 | first audit |
Questions
What is the NotebookLM Skill MCP server?
NotebookLM does the research, Claude writes the content. Research → Synthesis → Content Creation → Publishing. Claude Code Skill + MCP Server.
What tools does NotebookLM Skill expose?
13 in total: 7 read-only, 5 that write, and 1 that can delete or overwrite (nlm_delete). Every one is listed on this page with its risk.
Is NotebookLM Skill safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 1 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does NotebookLM Skill need?
It reads NOTEBOOKLM_AUTH_JSON from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does NotebookLM Skill run?
It speaks stdio, so it runs as a local process your client starts. It is published on PyPI as notebooklm-skill.
How current is this page?
The grade is for one exact copy of the source (addac501085e), read on 2026-10-01. The repository is watched and re-audited when it changes.