Gemini Skill

Overview

Use Google Gemini subscription for all AI and web search needs. No separate API keys required - uses your Google account.

Primary Tool: gemini CLI

The official Gemini CLI is installed at /opt/homebrew/bin/gemini.

Web Search (One-Shot)

gemini "What is the weather forecast for King, NC?"
gemini "Latest news about Epic EHR"
gemini "Research competitive landscape for X"

JSON Output (for scripting)

gemini "your query" --output-format json

Interactive Mode

gemini -i "Start researching topic X"

NotebookLM Integration

For deep research with source synthesis:

# Quick search (adds sources to notebook)
nlm source add-research "your query" --notebook <id>

# Deep research (adds sources to notebook)
nlm source add-research "your query" --notebook <id> --mode deep

# Query notebook sources
nlm ask -n <id> --new "your question"

When to Use What

Task	Tool
Quick web search	gemini "query"
One-shot Q&A	gemini "question"
Deep research with sources	nlm source add-research --mode deep
Query documents/sources	nlm ask -n <id>
Generate podcast	nlm generate audio

No API Keys Needed

Your Gemini subscription handles:

• Web search
• Text generation
• NotebookLM features
• All Gemini models

Do NOT use:

• Brave Search API (unnecessary cost)
• Separate Gemini API keys
• Other web search services

Authentication

Both tools use your Google account:

• gemini: Auto-authenticates via cached credentials
• nlm: Requires nlm login (browser-based, ~20 min sessions)

Files

• System CLI: /opt/homebrew/bin/gemini
• NotebookLM CLI: /opt/homebrew/bin/nlm
• Custom wrapper: skills/gemini/gemini_cli.py (optional, wraps both)