Anything To Notebooklm

PassAudited by VirusTotal on May 11, 2026.

Overview

Type: OpenClaw Skill Name: anything-to-notebooklm Version: 1.0.0 The 'anything-to-notebooklm' skill bundle is a legitimate tool designed to automate the processing of diverse content sources (WeChat, YouTube, Office documents, PDFs, etc.) and their integration with Google NotebookLM. It utilizes the Microsoft 'markitdown' library for file conversion and an external MCP server for WeChat scraping. The installation script (install.sh) and environment checker (check_env.py) are well-structured and transparent, and while they install dependencies from GitHub repositories, these actions are consistent with the stated functionality. No evidence of data exfiltration, malicious persistence, or harmful prompt injection was found; the instructions in SKILL.md are strictly focused on functional task execution.

Findings (0)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

The code actually run during installation or later MCP use could change if those repositories or their dependencies change or are compromised.

Why it was flagged

The installer pulls executable helper and CLI code from live GitHub repositories without a pinned commit, version, or hash; the cloned MCP server source is not included in the reviewed manifest.

Skill content
git clone https://github.com/Bwkyd/wexin-read-mcp.git "$MCP_DIR" ... pip3 install git+https://github.com/teng-lin/notebooklm-py.git -q
Recommendation

Pin external repositories to audited commits, provide lockfiles or vendored reviewed helper code, and install inside an isolated virtual environment before configuring Claude to use it.

What this means

A persistent helper expands the agent’s tool surface beyond this one prompt, and its actual behavior depends on code that is downloaded during setup rather than included in the reviewed package.

Why it was flagged

The instructions register a persistent MCP server in Claude’s configuration; after restart, that externally cloned server remains available to the agent.

Skill content
"mcpServers": { "weixin-reader": { "command": "python", "args": [ "/Users/joe/.claude/skills/anything-to-notebooklm/wexin-read-mcp/src/server.py" ] } } ... 配置后需要重启 Claude Code。
Recommendation

Only enable the MCP server if needed, review the cloned server code first, use an absolute path you control, and remove the MCP config entry when you no longer need WeChat article fetching.

What this means

This may violate site terms or trigger account/platform restrictions even though it is disclosed and related to the skill’s WeChat-reading purpose.

Why it was flagged

The skill openly states that the WeChat integration uses browser automation to bypass anti-crawling measures.

Skill content
微信公众号(绕过反爬虫) ... MCP 用浏览器模拟绕过。
Recommendation

Use the WeChat scraping feature only where you have permission and are comfortable with the platform-policy risk.

What this means

The NotebookLM CLI will operate with the user’s logged-in account and may create notebooks, add sources, and generate artifacts there.

Why it was flagged

The skill requires an authenticated NotebookLM account, which is expected for uploading and generating NotebookLM content but is not reflected in the registry credential declarations.

Skill content
首次使用前必须认证:

notebooklm login
notebooklm list  # 验证认证成功
Recommendation

Confirm which Google/NotebookLM account is logged in, understand the CLI’s permissions, and log out or use a separate account if needed.

What this means

Any selected local document or extracted ZIP content may be sent to Google NotebookLM and stored or processed there.

Why it was flagged

The workflow uploads converted sources to NotebookLM; supported inputs include local PDFs, Office files, images, audio, ZIP archives, webpages, and search summaries.

Skill content
notebooklm source add /tmp/weixin_xxx.txt --wait  # 上传文件并等待处理完成
Recommendation

Review files before use, avoid confidential or regulated content unless NotebookLM is approved for it, and delete notebooks/sources after processing if retention is not desired.