Twitter Curator

🤖 AI-powered Twitter automation system with persona-driven content generation and intelligent engagement.

Features

• ✅ Automated Tweet Posting - Schedule and post original tweets automatically
• ✅ URL Tracking - Capture and record tweet URLs for analytics
• ✅ Persona-Driven Content - Generate authentic content based on your personal profile
• ✅ Smart Engagement - Auto-reply to relevant tweets in your niche
• ✅ Anti-Detection - Uses Puppeteer with stealth plugin to avoid platform detection
• ✅ macOS Integration - LaunchAgents for scheduled automation

Tech Stack

• Node.js - Runtime environment
• Puppeteer v24.27.0 - Browser automation (M2 Mac compatible)
• Puppeteer Extra + Stealth Plugin - Anti-detection
• Google Gemini API - AI content generation
• macOS LaunchAgents - Scheduling system

Prerequisites

• Node.js 16+
• macOS (for LaunchAgents)
• Google Gemini API key
• Twitter account (logged in via Chrome profile)

Installation

# Clone the repository
git clone https://github.com/lmanchu/twitter-curator.git
cd twitter-curator

# Install dependencies
npm install

# Copy environment template
cp .env.example .env

# Edit .env and add your API keys
nano .env

Configuration

1. Environment Variables

Create a .env file:

GEMINI_API_KEY=your_gemini_api_key_here
HEADLESS=true
DRY_RUN=false

2. Persona Profile

Update PERSONA_FILE path in config.js to point to your persona profile markdown file.

3. Chrome Login

Run once with HEADLESS=false to login to Twitter:

HEADLESS=false node twitter-curator.js

Login manually when browser opens. Session will be saved to chrome-user-data/.

Usage

Manual Run

# Dry run (test without posting)
DRY_RUN=true node twitter-curator.js

# Live posting
node twitter-curator.js

Project Structure

twitter-curator/
├── twitter-curator.js        # Main automation script
├── content-generator.js      # AI content generation
├── config.js                 # Configuration
├── package.json              # Dependencies
├── .env                      # Environment variables (not committed)
├── chrome-user-data/         # Saved browser session (not committed)
├── logs/                     # Execution logs
└── posted-tweets.json        # Tweet history with URLs

How It Works

1. Content Generation: Uses Gemini AI to generate original tweets based on your persona
2. Browser Automation: Puppeteer navigates to Twitter and posts tweets
3. URL Capture: After posting, retrieves the tweet URL from your profile
4. Engagement: Searches for relevant tweets and generates contextual replies
5. Logging: Records all activities with timestamps and URLs

Key Technical Details

Button Selector

Uses [data-testid="tweetButtonInline"] for main tweet composition (not the modal tweetButton).

M2 Mac Compatibility

Puppeteer v24.27.0 resolves Rosetta compatibility issues on Apple Silicon.

Anti-Detection

• Stealth plugin to hide automation markers
• Random delays to mimic human behavior
• Persistent Chrome profile (no repeated logins)
• Custom user agent

Troubleshooting

"Post button not found"

• Check if Twitter UI has changed
• Try updating the selector in twitter-curator.js

Rosetta errors on M2 Mac

• Ensure Puppeteer is v24+: npm install puppeteer@^24.27.0

Login required every time

• Check chrome-user-data/ directory exists
• Run once with HEADLESS=false to login

Version History

See CHANGELOG.md for detailed version history.

License

MIT

Disclaimer

⚠️ Use responsibly: Automated posting may violate Twitter's Terms of Service. This tool is for educational purposes. Use at your own risk.

Author

Built with ❤️ by Lman

---

🤖 Generated with Claude Code

Co-Authored-By: Claude noreply@anthropic.com