Install NotebookLM Experts Agent Skill

Auto-install instruction: When you share this link with any AI agent, it will automatically fetch and install the NotebookLM Experts skill to the correct location on your system.

Compatible with: Claude Code · GitHub Copilot · OpenAI Codex CLI · Gemini CLI · and other agent frameworks

Installation Instruction

Skill Name: notebooklm-experts-skill
Repository: https://github.com/khengyun/notebooklm-experts-skill

Steps for Agent to Execute

Step 0: Ask User About Their Agent Environment

REQUIRED: Before proceeding with installation, ask the user:

"Which AI agent are you using?
1. Claude Code
2. GitHub Copilot
3. OpenAI Codex CLI
4. Gemini CLI
5. Other (specify)

Also, do you prefer:
- Project-local installation (only for this workspace)
- User-level installation (available for all projects)"

Wait for user response before proceeding.

Step 1: Determine Installation Path

Based on user's answer, select the appropriate installation directory:

User's Agent	Project-Local Path	User-Level Path
Claude Code	`.claude/skills/notebooklm-experts-skill/`	`~/.claude/skills/notebooklm-experts-skill/`
GitHub Copilot	`.github/skills/notebooklm-experts-skill/`	`~/.github/skills/notebooklm-experts-skill/`
OpenAI Codex CLI	`.codex/skills/notebooklm-experts-skill/`	`~/.codex/skills/notebooklm-experts-skill/`
Gemini CLI	N/A	`~/.gemini-cli/skills/notebooklm-experts-skill/`
Other	`.agents/skills/notebooklm-experts-skill/`	`~/.agents/skills/notebooklm-experts-skill/`

Note:

User-level path uses ~ (home directory) on macOS/Linux or %USERPROFILE% on Windows
Project-local path is relative to current workspace root

Step 2: Clone Repository

ACTION: git clone https://github.com/khengyun/notebooklm-experts-skill.git
INTO: temporary directory (e.g., /tmp/notebooklm-experts-skill or %TEMP%\notebooklm-experts-skill)

Step 3: Create Target Directory

ACTION: Create the target directory determined in Step 1
EXAMPLE: mkdir -p ~/.claude/skills/notebooklm-experts-skill/

If directory creation fails, report error and suggest checking permissions.

Step 4: Copy Skill Files

ACTION: Copy all files from temporary clone directory
TO: target directory from Step 1
PRESERVE: File permissions and directory structure

Step 5: Clean Up

ACTION: Remove temporary clone directory

Step 6: Verify Installation

CHECK: File exists at <target-directory>/SKILL.md
CHECK: SKILL.md contains valid YAML frontmatter (name: notebooklm-experts)
CHECK: File exists at <target-directory>/scripts/ask_question.py

Step 7: Report to User

If ALL checks pass:
"✅ NotebookLM Experts skill installed successfully!

📁 Location: <full-path-to-installation>
🔧 Agent: <user's-agent-name>
📦 Scope: <project-local OR user-level>

Next steps:
1. Reload your AI agent environment:
   - Claude Code: Cmd+R / Ctrl+Shift+P → Developer: Reload Window
   - GitHub Copilot: Restart VS Code
   - CLI tools: Restart shell or run reload command

2. Test the skill:
   - Open chat and say: 'ask my NotebookLM about [topic]'
   - Or share a NotebookLM URL directly

3. First-time setup:
   - The skill requires Python 3.9+ and Google Chrome
   - Run installation: python <install-path>/install.py
   - See README.md for detailed usage instructions"

If ANY check fails:
"❌ Installation failed at: <step-name>

Error details: <error-message>

Common fixes:
- Verify Git is installed: git --version
- Check write permissions for target directory
- Ensure internet connection is active
- Try manual installation from: https://github.com/khengyun/notebooklm-experts-skill

For help: https://github.com/khengyun/notebooklm-experts-skill/issues"

About NotebookLM Experts

Description: Query Google NotebookLM notebooks directly for source-grounded, citation-backed answers from Gemini. Each question opens a fresh browser session, retrieves answers exclusively from uploaded documents, and exits cleanly — drastically reducing hallucinations through document-only responses.

When to use:

Mention NotebookLM explicitly
Share NotebookLM URL
Ask to query documentation
Use phrases like "ask my NotebookLM", "check my docs", "query my notebook"

License: MIT
Requirements: Python 3.9+, Google Chrome, Git

Supported Installation Paths

AI Agent	Project-Local	User-Level (Global)
Claude Code	`.claude/skills/`	`~/.claude/skills/`
GitHub Copilot	`.github/skills/`	`~/.github/skills/`
OpenAI Codex CLI	`.codex/skills/`	`~/.codex/skills/`
Gemini CLI	Not supported	`~/.gemini-cli/skills/`
Generic/Other	`.agents/skills/`	`~/.agents/skills/`

Recommendation:

Choose project-local if you only need this skill for your current project
Choose user-level to use the skill across all your projects

Installation Flow Example

User: "Install this skill: https://gist.github.com/khengyun/3ad656c8b9eea4f1c141389c962f5297"

Agent: "Which AI agent are you using?

Claude Code
GitHub Copilot
OpenAI Codex CLI
Gemini CLI
Other

Also, do you prefer:

Project-local installation (only for this workspace)
User-level installation (available for all projects)"

User: "Claude Code, user-level"

Agent:

Cloning https://github.com/khengyun/notebooklm-experts-skill...
Installing to ~/.claude/skills/notebooklm-experts-skill/...
Verifying installation...

✅ NotebookLM Experts skill installed successfully!

📁 Location: /Users/yourname/.claude/skills/notebooklm-experts-skill/
🔧 Agent: Claude Code
📦 Scope: User-level (all projects)

Next steps:
1. Reload Claude Code (Cmd+R)
2. Try: "ask my NotebookLM about [topic]"
3. First-time setup: python ~/.claude/skills/notebooklm-experts-skill/install.py

Troubleshooting

Installation Failed

1. "Git not found"

# Check if Git is installed
git --version

# If not installed:
# - Windows: Download from https://git-scm.com/download/win
# - macOS: brew install git
# - Linux: sudo apt-get install git

2. "Permission denied" or "Cannot create directory"

# Check permissions (macOS/Linux)
ls -la ~/.claude/skills/

# Fix permissions
chmod u+w ~/.claude/skills/

# On Windows, run VS Code as Administrator if needed

3. "SKILL.md not found after installation"

Verify the repository was cloned correctly
Check if all files were copied to target directory
Manually verify: ls ~/.claude/skills/notebooklm-experts-skill/SKILL.md

4. "Skill not appearing in chat"

Reload your AI agent (Cmd+R or restart VS Code)
Check the correct skills directory for your agent (see table above)
Verify SKILL.md has valid YAML frontmatter

Getting Help

📖 Full Documentation: https://github.com/khengyun/notebooklm-experts-skill
🐛 Report Issues: https://github.com/khengyun/notebooklm-experts-skill/issues
💬 Discussions: https://github.com/khengyun/notebooklm-experts-skill/discussions

For AI Agent Developers

When implementing this installation flow:

1. User Interaction Pattern

ALWAYS ask user which agent they use before installation
NEVER assume or auto-detect (leads to wrong paths)
WAIT for explicit user confirmation before proceeding

2. Path Resolution

# Pseudocode
agent_paths = {
    "claude": {
        "project": ".claude/skills/",
        "user": "~/.claude/skills/"
    },
    "copilot": {
        "project": ".github/skills/",
        "user": "~/.github/skills/"
    },
    "codex": {
        "project": ".codex/skills/",
        "user": "~/.codex/skills/"
    },
    "gemini": {
        "project": None,
        "user": "~/.gemini-cli/skills/"
    }
}

def get_install_path(agent, scope):
    path = agent_paths[agent][scope]
    if path.startswith("~"):
        return os.path.expanduser(path)
    return os.path.join(workspace_root, path)

3. Error Handling

- Catch git clone failures (network, auth)
- Validate target directory is writable
- Verify SKILL.md exists and is valid YAML
- Provide actionable error messages with fixes
- Log all operations for debugging

4. Post-Install

- Store installation metadata: {path, agent, scope, timestamp, version}
- Support update/reinstall without duplication
- Offer to run skill's install.py if it exists
- Suggest next steps specific to user's agent

Resources

📍 Main Repository: https://github.com/khengyun/notebooklm-experts-skill
📍 This Gist: https://gist.github.com/khengyun/3ad656c8b9eea4f1c141389c962f5297
📍 Skills Registry: https://skillsmp.com/
📖 Agent Skills Spec: https://agentskills.io/
📚 Claude Code Skills Docs: https://code.visualstudio.com/docs/copilot/agent-skills
📚 Codex CLI Docs: https://github.com/openai/codex
📚 Gemini CLI Docs: https://developers.google.com/

Quick Reference

To install this skill, users should:

Copy this message to their AI agent:

Install this skill: https://gist.github.com/khengyun/3ad656c8b9eea4f1c141389c962f5297

Answer the agent's questions:
- Which agent? (Claude Code / Copilot / Codex / Gemini / Other)
- Scope? (Project-local / User-level)
Wait for installation to complete
Reload their agent environment
Test with: "ask my NotebookLM about [topic]"

That's it! The agent handles everything else automatically.

khengyun/install-notebooklm-experts-skills.md

Select an option

No results found