Skip to content

README.md

Latest commit

 

History

History
245 lines (193 loc) · 6.66 KB

README.md

File metadata and controls

245 lines (193 loc) · 6.66 KB

CCAI Email Automation Setup Guide

1. Install Dependencies

Basic dependencies:

pip3 install aiohttp python-dotenv pytz

For LinkedIn scraping functionality:

pip3 install linkedin_scraper

For AWS Bedrock AI email generation:

pip3 install boto3

Using a virtual environment (recommended):

python3 -m venv path/to/venv
source path/to/venv/bin/activate
python3 -m pip install aiohttp python-dotenv pytz linkedin_scraper boto3

2. Install ChromeDriver (for LinkedIn scraping)

macOS (using Homebrew):

brew install chromedriver

Alternative installation methods:

  • Download from ChromeDriver official site
  • Ensure ChromeDriver version matches your Chrome browser version
  • Add ChromeDriver to your PATH

Verify installation:

chromedriver --version

3. Configure Environment File

Create a .env file in the project root with your CCAI credentials:

CCAI_API_KEY=your_api_key_here
CCAI_CLIENT_ID=your_client_id_here
CCAI_EMAIL_URL=your_email_endpoint_url_here

# Sender Information (used in email signatures)
SENDER_NAME=Your Name
SENDER_TITLE=Your Title
SENDER_EMAIL=your.email@company.com
SENDER_PHONE=+1234567890
SENDER_LINKEDIN=https://www.linkedin.com/in/yourprofile
SENDER_COMPANY=Your Company Name
SENDER_COMPANY_URL=https://yourcompany.com

# AWS Bedrock settings
AWS_REGION=us-east-1

4. Configure AWS Bedrock (for AI email generation)

Option 1: AWS SSO/IAM Identity Center (Recommended for Organizations)

# Install AWS CLI if not already installed
brew install awscli

# Configure SSO
aws configure sso

You'll be prompted for:

  • SSO session name (e.g., "my-company")
  • SSO start URL (e.g., https://my-company.awsapps.com/start)
  • SSO region (e.g., us-east-1)
  • Account ID and Role name
  • Default region (us-east-1) and output format (json)
# Login to SSO (opens browser)
aws sso login --profile your-profile-name

# Or if set as default profile:
aws sso login

Option 2: AWS CLI Configuration (Individual Users)

# Install AWS CLI
brew install awscli

# Configure credentials
aws configure

Enter your AWS Access Key ID, Secret Access Key, region (us-east-1), and output format (json).

Option 3: Environment Variables

Add to your .env file:

AWS_ACCESS_KEY_ID=your_access_key_here
AWS_SECRET_ACCESS_KEY=your_secret_key_here
AWS_REGION=us-east-1

# Optional: Specify AWS profile for SSO
AWS_PROFILE=your-sso-profile-name

AWS Credential Priority Order:

The scripts will attempt to use AWS credentials in this order:

  1. AWS Profile (if AWS_PROFILE is set) - SSO profiles supported
  2. Environment Variables (if profile fails or not set) - AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY
  3. Default AWS Credentials - IAM roles, default profile, or AWS CLI configuration

Enable Bedrock Model Access:

  1. Go to AWS Bedrock Console
  2. Navigate to "Model access"
  3. Enable Amazon Nova Pro model (us.amazon.nova-pro-v1:0)
  4. Submit the request (approval may take a few minutes)

5. LinkedIn Authentication Setup

Option 1: Session Cookies (Recommended)

  1. Log into LinkedIn in your browser
  2. Open Developer Tools (F12)
  3. Go to Application/Storage → Cookies → linkedin.com
  4. Copy the li_at cookie value
  5. Add to your .env file:
LINKEDIN_SESSION_COOKIE=your_li_at_cookie_value_here

Option 2: Username/Password (Less secure)

Add to your .env file:

LINKEDIN_USERNAME=your_linkedin_email@example.com
LINKEDIN_PASSWORD=your_linkedin_password

Note: Session cookies are more secure and less likely to trigger LinkedIn's security measures.

6. Enable WebDriver (if disabled)

If you see "WebDriver disabled" messages, re-enable it by editing: linkedin_mcp_server/error_handler.py

Replace the safe_get_driver() function with:

def safe_get_driver():
    """
    Safely get or create a driver with proper error handling.
    """
    from linkedin_mcp_server.authentication import ensure_authentication
    from linkedin_mcp_server.drivers.chrome import get_or_create_driver

    # Get authentication first
    authentication = ensure_authentication()

    # Create driver with authentication
    driver = get_or_create_driver(authentication)

    return driver

7. Usage Examples

Send a single email:

python3 send_single_email.py "recipient@email.com" "Subject" "Message" 30

Run LinkedIn email automation:

python3 ai_gmail_outbound.py

Run test email campaign with AI generation:

python3 send_test_emails.py

Preview AI email templates:

python3 simple_test_emails.py

Note: All scripts now support AWS profiles for Bedrock AI generation. They will automatically use the AWS_PROFILE environment variable if set, or fall back to default AWS credentials.

8. Troubleshooting

Common Issues:

ChromeDriver not found:

brew install chromedriver
# or download manually and add to PATH

AWS Bedrock access denied:

  • Ensure you have proper AWS credentials configured
  • For SSO: Run aws sso login --profile your-profile-name to refresh session
  • Check that Amazon Nova Pro model is enabled in Bedrock console
  • Verify your AWS region is set to us-east-1
  • For SSO profiles, ensure the role has Bedrock permissions

AWS SSO session expired:

# Refresh your SSO session
aws sso login --profile your-profile-name

# Check current session status
aws sts get-caller-identity --profile your-profile-name

LinkedIn login issues:

  • Use session cookies instead of username/password
  • Check if your LinkedIn account has 2FA enabled
  • Ensure cookies are fresh (re-login if needed)

Module not found errors:

  • Ensure you're in the virtual environment: source path/to/venv/bin/activate
  • Install missing dependencies: pip install <package-name>

Rate limiting:

  • LinkedIn may rate limit requests
  • Add delays between profile scraping
  • Use session cookies for better authentication

Dependencies Overview

  • aiohttp: For async HTTP requests to CCAI API
  • python-dotenv: For loading environment variables from .env file
  • pytz: For timezone handling in email scheduling
  • linkedin_scraper: For LinkedIn profile and post scraping
  • selenium: WebDriver automation (installed with linkedin_scraper)
  • boto3: AWS SDK for Bedrock AI email generation

Security Notes

  • Never commit your .env file to version control
  • Use session cookies instead of passwords when possible
  • Regularly refresh LinkedIn session cookies
  • Store AWS credentials securely (use IAM roles when possible)
  • Be mindful of LinkedIn's rate limits and terms of service