CrewAI Blog Generation System

🚀 Generate high-quality blog posts on any topic using AI agents!

This system uses 6 specialized AI agents to automatically research, write, optimize, and publish comprehensive blog posts. Perfect for content creators, bloggers, and businesses who need consistent, high-quality content.

✨ What It Does

• 🔍 Researches any topic using real-time web search
• ✍️ Writes comprehensive 3500-word articles with proper structure
• 🎯 Optimizes content for SEO with auto-generated tags
• 🌐 Formats content as clean HTML for Ghost CMS
• ✅ Reviews quality and accuracy before publication
• 📝 Publishes directly to Ghost CMS as drafts (automatic posting!)

⚠️ Requirements

Python 3.10 or higher required (Python 3.12 recommended)

CrewAI 0.5.0 uses modern type hints (Type | None) that require Python 3.10+. If you're using Python 3.9 or earlier, you'll need to upgrade.

🚀 Quick Start

1. Check Your Python Version

python3.12 --version  # Should show 3.12.x

Don't have Python 3.12? Install it:

• macOS (Homebrew): brew install [email protected]
• Ubuntu/Debian: sudo apt install python3.12 python3.12-venv
• Windows: Download from python.org

2. Setup (5 minutes)

# Clone the repository
git clone https://git.christianmendieta.ca/christancho/Blogging-with-CrewAI.git
cd Blogging-with-CrewAI

# Create virtual environment with Python 3.12 (recommended)
python3.12 -m venv venv
source venv/bin/activate  # On macOS/Linux
# OR: venv\Scripts\activate  # On Windows

# Install dependencies
pip install --upgrade pip
pip install -r requirements.txt

# Set up your API keys
cp .env.example .env

3. Get Your API Keys

Required:

• OpenAI API Key: Get it here
• Brave Search API Key: Get it here (2000 free searches/month)
• Ghost CMS API Key: Create in your Ghost admin panel
• Ghost CMS URL: Your Ghost site URL (e.g., https://your-site.com)

Getting Your Ghost CMS API Key

1. Log into your Ghost admin panel (usually https://your-site.com/ghost)
2. Go to Settings → Integrations
3. Click "Add custom integration"
4. Give it a name (e.g., "Blog Generator")
5. Copy the Admin API Key (starts with something like 5d4c...) - NOT the Content API Key
6. Use your Ghost site URL (e.g., https://your-site.com)

⚠️ Important: You need the Admin API Key, not the Content API Key, to create posts!

4. Configure Your Keys

Edit your .env file:

# Required
OPENAI_API_KEY=sk-your-openai-key-here
BRAVE_SEARCH_API_KEY=your-brave-search-key-here
GHOST_API_KEY=your-ghost-api-key-here
GHOST_API_URL=https://your-ghost-site.com

5. Generate Your First Blog Post

# Interactive mode (recommended for first time)
python main.py

# Or specify a topic directly
python main.py --topic "How to start a successful blog"

💡 LLM Provider Configuration

The system now supports ANY LLM provider! Configure your preferred provider (OpenAI, Anthropic Claude, OpenRouter, local models, etc.)

Quick Setup - OpenAI (Default):

LLM_API_BASE_URL=https://api.openai.com/v1
LLM_API_KEY=your-openai-api-key
LLM_MODEL_NAME=gpt-4o-mini
LLM_TEMPERATURE=0.7

🌟 Supported Providers

OpenAI (Recommended)

LLM_API_BASE_URL=https://api.openai.com/v1
LLM_API_KEY=sk-your-key-here
LLM_MODEL_NAME=gpt-4o-mini  # 128K context, fast, cheap
LLM_TEMPERATURE=0.7

Anthropic Claude

LLM_API_BASE_URL=https://api.anthropic.com/v1
LLM_API_KEY=sk-ant-your-key-here
LLM_MODEL_NAME=claude-3-5-sonnet-20241022  # 200K context
LLM_TEMPERATURE=0.7

OpenRouter (Access to Multiple Models)

LLM_API_BASE_URL=https://openrouter.ai/api/v1
LLM_API_KEY=sk-or-your-key-here
LLM_MODEL_NAME=anthropic/claude-3-5-sonnet  # or openai/gpt-4o-mini
LLM_TEMPERATURE=0.7

Local Models (Ollama)

LLM_API_BASE_URL=http://localhost:11434/v1
LLM_API_KEY=not-needed
LLM_MODEL_NAME=llama3:70b  # or mixtral:8x7b
LLM_TEMPERATURE=0.7

⚠️ Important Requirements

Your model MUST have at least 128K context window to avoid errors. Models with smaller context (like gpt-4 with 8K) will fail.

Recommended Models:

• ✅ gpt-4o-mini (OpenAI, 128K, fast, cheap)
• ✅ gpt-4o (OpenAI, 128K, most capable)
• ✅ claude-3-5-sonnet-20241022 (Anthropic, 200K)
• ✅ llama3:70b (Local, 128K+ via Ollama)

Avoid These:

• ❌ gpt-4 - Only 8K context
• ❌ gpt-3.5-turbo - Only 16K context
• ❌ Small local models with <128K context

📝 How It Works

The system uses 6 AI agents working together:

1. 🔍 Research Agent    → Finds current information about your topic
2. ✍️ Content Writer    → Creates a 3500-word structured article  
3. 🎯 SEO Optimizer     → Optimizes for search engines + generates tags
4. 🌐 HTML Formatter    → Converts to Ghost CMS-ready HTML
5. ✅ Quality Reviewer   → Checks accuracy and quality
6. 📝 Ghost Publisher   → Posts to Ghost CMS as draft (automatic!)

🎯 Example Topics

The system works with any topic - not just technical ones:

Business & Marketing:

• "Digital marketing strategies for small businesses"
• "How to build a successful startup"

Lifestyle & Health:

• "Sustainable living practices for urban dwellers"
• "Nutrition strategies for busy professionals"

Technology:

• "Python virtual environments best practices"
• "Docker containerization guide"

Education:

• "Effective online learning techniques"
• "Teaching coding to beginners"

And many more! Just provide any topic and the AI will research and write about it.

📊 Article Structure

Every generated article follows this structure:

• Introduction (400-500 words) - Hook, context, overview
• Section 1 (600-700 words) - Foundational concepts
• Section 2 (600-700 words) - Detailed information
• Section 3 (600-700 words) - Advanced concepts & best practices
• Section 4 (500-600 words) - Future trends & actionable insights
• Conclusion (300-400 words) - Key takeaways & next steps

Total: ~3500 words with SEO optimization and auto-generated tags

🛠️ Usage Examples

Basic Usage

# Interactive mode - system will ask for your topic
python main.py

# Direct topic specification
python main.py --topic "Remote work productivity tips"

# Skip approval step (auto-approve)
python main.py --topic "Climate change solutions" --no-approval

# List all generated posts
python main.py --list

Advanced Usage

# Generate multiple posts
python main.py --topic "AI in healthcare"
python main.py --topic "Sustainable energy solutions"
python main.py --topic "Digital nomad lifestyle"

📁 File Structure

Blogging-with-CrewAI/
├── main.py              # Main script - run this!
├── agents.py            # AI agent definitions
├── tasks.py             # What each agent does
├── tools.py             # Search, SEO, and publishing tools
├── config.py            # Configuration settings
├── requirements.txt     # Python dependencies
├── .env                 # Your API keys (create this)
└── output/              # Generated blog posts (auto-created)

🔧 Configuration Options

You can customize the system by editing your .env file:

# LLM Provider (supports ANY provider!)
LLM_API_BASE_URL=https://api.openai.com/v1
LLM_API_KEY=your-key-here
LLM_MODEL_NAME=gpt-4o-mini
LLM_TEMPERATURE=0.7

# Search API
BRAVE_SEARCH_API_KEY=your-key-here

# Ghost CMS (for publishing)
GHOST_API_KEY=your-key-here
GHOST_API_URL=https://your-site.com
GHOST_AUTHOR_ID=1

# Legacy OpenAI vars (deprecated, use LLM_* above)
OPENAI_API_KEY=your-key-here
OPENAI_MODEL_NAME=gpt-4o-mini
OPENAI_TEMPERATURE=0.7

Example Configurations

Using Anthropic Claude:

LLM_API_BASE_URL=https://api.anthropic.com/v1
LLM_API_KEY=sk-ant-your-key
LLM_MODEL_NAME=claude-3-5-sonnet-20241022
LLM_TEMPERATURE=0.7

Using Local Ollama:

LLM_API_BASE_URL=http://localhost:11434/v1
LLM_API_KEY=not-needed
LLM_MODEL_NAME=llama3:70b
LLM_TEMPERATURE=0.7

🚨 Troubleshooting

Common Issues

"Configuration Error: Missing required environment variables"

• Make sure your .env file exists and has valid API keys
• Check that all required keys are set: OPENAI_API_KEY, BRAVE_SEARCH_API_KEY, GHOST_API_KEY, GHOST_API_URL

TypeError: unsupported operand type(s) for |

• Cause: You're using Python 3.9 or earlier
• Solution: Use Python 3.10 or higher (preferably 3.12)

# Check your Python version
python --version

# If it's < 3.10, recreate venv with Python 3.12
rm -rf venv
python3.12 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

"Rate limit exceeded"

• The system now has a very high TPM limit (200,000) to reduce rate limiting
• Switch to gpt-4o-mini model for even better performance
• Wait a few minutes and try again if needed
• Check your OpenAI billing status

"Search failed with status code: 401"

• Verify your Brave Search API key is correct
• Check if you've exceeded your 2000 free searches/month

cannot import name 'BaseTool' from 'crewai.tools'

• Cause: Import issue (already fixed in latest version)
• Solution: Make sure you have the latest code with imports from langchain.tools

"Ghost CMS authentication failed"

• Make sure you're using the Admin API Key, not the Content API Key
• Verify your Ghost CMS URL is correct (should end with your domain, not /ghost)
• Check that your Ghost site is accessible

"Ghost CMS API endpoint not found"

• Ensure your Ghost CMS URL is correct (e.g., https://your-site.com)
• Make sure your Ghost site is running and accessible
• Check that you have the correct permissions on your Ghost site

"Error during blog generation"

• Check your OpenAI API key and billing status
• Ensure you have sufficient API credits
• Try with a simpler topic first

Warning: Mixing V1 models and V2 models

• This is safe to ignore - it's a pydantic compatibility warning from CrewAI 0.5.0 and won't affect functionality

Virtual Environment Activation

Every time you work on the project, remember to activate the virtual environment:

cd "/path/to/Blogging-with-CrewAI"
source venv/bin/activate  # macOS/Linux
# OR
venv\Scripts\activate     # Windows

When you're done: deactivate

Getting Help

1. Check this README - Most issues are covered here
2. Check your API keys - Make sure they're valid and have credits
3. Try GPT-4o-mini - Solves most rate limit issues
4. Start simple - Try a basic topic first to test the system

💰 Cost Estimates

Per blog post (3500 words):

| Provider | Cost per Post | Notes | |----------|---------------|-------| | OpenAI (gpt-4o-mini) | ~$0.10-0.20 | ✅ Recommended | | OpenAI (gpt-4o) | ~$0.50-1.00 | More capable but pricier | | Anthropic Claude | ~$0.15-0.30 | Great alternative | | OpenRouter | Varies | Pay-as-you-go from multiple providers | | Local (Ollama) | FREE | Requires local GPU/CPU resources | | Brave Search | FREE | 2000 searches/month included | | Ghost CMS | FREE | If you have Ghost hosting |

Most cost-effective setup: Local Ollama + Free Brave Search = $0.00 per post 🎉

🎯 What You Get

After running the system, you'll have:

1. 📄 HTML file in the output/ folder
2. 📝 Draft in Ghost CMS (automatically created!)
3. 🏷️ Auto-generated tags for SEO
4. 📊 SEO-optimized content ready for publication
5. ✅ Quality-reviewed article that's ready to publish

🚀 Next Steps

1. Generate your first post using the Quick Start guide
2. Check your Ghost CMS for the automatically created draft
3. Experiment with different topics to see the system's versatility
4. Customize the system by modifying agent prompts in agents.py

📈 Tips for Best Results

• Be specific with topics: "Python async programming" vs "programming"
• Use GPT-4o-mini to avoid rate limits
• Review the preview before approving publication
• Start with simpler topics to test the system
• Check your API credits before generating multiple posts

---

Ready to generate amazing blog content? Start with the Quick Start guide above! 🚀