MQL5-Google-Onedrive/docs/GEMINI_CLI_INTEGRATION.md

Gemini CLI Integration Guide

Overview

This project now includes Google Gemini CLI as a local npm dependency, providing AI capabilities for market analysis, trade confirmation, and code upgrades.

What's New

Version Integrated

• Package: @google/gemini-cli
• Version: 0.30.0-nightly.20260212.207ac6f2d
• Includes: Commit 6a2f2d3a91380985c9ada6d39219fb353b4291b2 with headless mode fixes

Key Improvements

1. Better Headless Mode Detection: Fixed detection for non-interactive environments
2. CI Environment Support: Automatic detection of CI/GitHub Actions environments
3. Query Option Support: Added support for query parameter
4. Command-line Flag Detection: Fallback checking for -p, --prompt, -y, and --yolo flags

Installation

First Time Setup

# Clone the repository
git clone https://github.com/A6-9V/MQL5-Google-Onedrive.git
cd MQL5-Google-Onedrive

# Install all dependencies including Gemini CLI
npm install

# Or use the setup script
bash setup.sh
# Then select option 2: Install npm dependencies

Verify Installation

# Check version
npx gemini --version

# Should output: 0.30.0-nightly.20260212.207ac6f2d

# View help
npx gemini --help

Usage

Interactive Mode (Terminal Chat)

npx gemini

This starts an interactive chat session with Gemini.

Non-Interactive Mode (Headless)

# Single prompt
npx gemini --prompt "Analyze the EURUSD market trend"

# With specific model
npx gemini --model gemini-2.0-flash --prompt "Review this trading strategy"

# YOLO mode (auto-approve all actions)
npx gemini --yolo --prompt "Generate a market report"

Integration with Trading System

The Gemini CLI is used in several places:

1. Trade Confirmation (via EA):

   • Set UseGeminiFilter = true
   • Set AiProvider = PROVIDER_GEMINI
   • Add your API key to GeminiApiKey

2. Market Research (automated scripts):

   python scripts/market_research.py
   

3. Code Upgrades (automated suggestions):

   python scripts/upgrade_repo.py
   

Configuration

API Key Setup

You need a Google AI API key from Google AI Studio.

Option 1: Environment Variable (Recommended)

Add to your shell profile or .env file:

export GEMINI_API_KEY="your_api_key_here"

Option 2: Settings File

Create ~/.gemini/settings.json:

{
  "auth": {
    "apiKey": "your_api_key_here"
  }
}

MT5 WebRequest Setup

For EA integration, enable WebRequests in MT5:

1. Open MT5 → Tools → Options → Expert Advisors
2. Check "Allow WebRequest for listed URL"
3. Add: https://generativelanguage.googleapis.com

Testing the Integration

Basic Test

# Test version
npx gemini --version

# Test non-interactive mode
npx gemini --prompt "Hello, Gemini!"

Full Setup Test

# Run full validation
bash setup.sh --ci

# Or interactive menu
bash setup.sh
# Select option 1: Run full validation

Troubleshooting

"gemini: command not found"

If you see this error, use npx:

npx gemini --version

Or install globally:

npm install -g @google/gemini-cli

API Key Issues

If you get authentication errors:

1. Verify your API key at Google AI Studio
2. Check the key is set in environment or settings file
3. Test with: npx gemini --prompt "test"

WebRequest Errors in MT5

If the EA can't connect to Gemini:

1. Check MT5 allowed URLs (Tools → Options → Expert Advisors)
2. Add https://generativelanguage.googleapis.com
3. Restart MT5 terminal

Features Enabled by Gemini CLI

1. AI Trade Confirmation

The EA can ask Gemini to confirm trades before execution:

// In EA settings
input bool UseGeminiFilter = true;
input ENUM_AI_PROVIDER AiProvider = PROVIDER_GEMINI;
input string GeminiApiKey = "your_key_here";

2. Market Research Automation

Automated market analysis reports:

./scripts/setup_research.sh  # First-time setup
python scripts/market_research.py  # Run analysis

Generates: docs/market_research_report.md

3. Code Upgrade Suggestions

AI-powered code improvement suggestions:

python scripts/upgrade_repo.py

Generates: docs/upgrade_suggestions.md

4. Scheduled Research

Automatic research every 4 hours:

python scripts/schedule_research.py

Advanced Usage

Custom Models

# Use specific model
npx gemini --model gemini-2.0-flash-exp --prompt "Analyze..."

# Use preview model
npx gemini --model gemini-1.5-pro-preview --prompt "Analyze..."

Approval Modes

# Default mode (prompt for approval)
npx gemini --prompt "Analyze..."

# Auto-edit mode (auto-approve edit tools)
npx gemini --approval-mode auto_edit --prompt "Fix..."

# YOLO mode (auto-approve all)
npx gemini --approval-mode yolo --prompt "Do everything..."

# Plan mode (read-only)
npx gemini --approval-mode plan --prompt "Plan..."

Session Management

# List sessions
npx gemini --list-sessions

# Resume session
npx gemini --resume latest

# Delete session
npx gemini --delete-session 5

Security Notes

1. Never commit API keys to the repository
2. Use .env file for local development (already in .gitignore)
3. Use GitHub Secrets for CI/CD workflows
4. Rotate API keys regularly
5. Monitor API usage at Google AI Studio

Resources

• Gemini CLI Documentation
• Google AI Studio
• Gemini API Documentation
• Local Setup Guide

Support

For issues with:

• Gemini CLI itself: https://github.com/google-gemini/gemini-cli/issues
• This integration: Open an issue in this repository
• MT5 EA integration: See EA documentation