Content
# CrewAI MCP Demo
[](https://www.python.org/downloads/) [](https://opensource.org/licenses/MIT) [](https://github.com/crewai/crewai)
This repository demonstrates **different approaches** for using CrewAI with MCP (Model Context Protocol) to interact with external tools and services through a standardized protocol.
## 🌟 **Featured Examples**
### 🏭 **Enterprise Regulatory Intelligence**
**📍 Location:** `scaffolding_approach_examples/snowflake_mcp_demo/`
A **production-ready 3-agent system** that combines CrewAI's multi-agent orchestration with Snowflake's enterprise data capabilities to build regulatory intelligence systems.
**Key Features:**
- 🤖 **3-Agent Coordination**: Regulatory Intelligence + SEC Filing Analysis + Market News
- 📊 **Enterprise Scale**: Semantic search across 372,299 SEC filings via Snowflake MCP
- 🎯 **Professional Output**: Investment-grade regulatory impact reports
- 🚀 **Production Ready**: Complete deployment and configuration examples
[**→ View Detailed Tutorial**](scaffolding_approach_examples/snowflake_mcp_demo/)
---
## 🚀 **Two Integration Approaches**
This repository demonstrates **two different ways** to integrate CrewAI with MCP:
### 1. 🏗️ **Scaffolding Approach** (Recommended for Production)
- Uses CrewAI's project scaffolding with `@CrewBase` decorator
- Structured project with configuration files
- Built-in `get_mcp_tools()` method (CrewAI 0.134.0+)
- Better for complex, maintainable applications
**Examples:**
- **Snowflake Enterprise Demo** - Regulatory intelligence system
- **Mathematician Project** - Math operations via local StdIO server
- **Context7 Integration** - Document search via streamable-HTTP
### 2. 📝 **Script Approach** (Great for Learning/Prototyping)
- Standalone Python scripts using `MCPServerAdapter` directly
- Quick to set up and experiment with
- Perfect for understanding MCP concepts
**Examples:**
- `stdio_client_demo.py` - Math operations
- `sse_client_demo.py` - Cloudflare docs search
- `streamable_http_client_demo.py` - Simple greeting
- `multiple_servers_client_demo.py` - Combined functionality
---
## 🔧 **Transport Mechanisms**
### **Streamable HTTP** ⭐ (Recommended)
- **Use case**: Remote hosted servers (Context7, cloud services)
- **Benefits**: Simple setup, bidirectional communication, web-friendly
- **Example**: Snowflake MCP, Context7 API
### **Server-Sent Events (SSE)**
- **Use case**: Real-time server-to-client updates
- **Benefits**: Good for streaming responses, widely supported
- **Example**: Cloudflare docs, real-time data feeds
### **Standard Input/Output (StdIO)**
- **Use case**: Local development, testing, process-based servers
- **Benefits**: Simple local setup, no network required
- **Example**: Local math server, file processing tools
---
## 📁 **Project Structure**
```
crewai-mcp-demo/
├── 🏗️ scaffolding_approach_examples/
│ ├── snowflake_mcp_demo/ # 🌟 FEATURED: Enterprise regulatory intelligence
│ ├── mathematician_project/ # Math operations via local StdIO server
│ └── crewai_context7_mcp/ # Context7 integration via streamable-HTTP
├── 📝 script_approach_examples/ # Standalone script examples
│ ├── stdio_client_demo.py # Math operations via StdIO
│ ├── sse_client_demo.py # Cloudflare docs via SSE
│ ├── streamable_http_client_demo.py # Greeting via HTTP
│ └── multiple_servers_client_demo.py # Multiple servers example
├── 🖥️ servers/ # Local MCP servers
│ ├── hello_http_server.py # HTTP greeting server
│ └── math_stdio_server.py # StdIO math server
└── README.md # This file
```
---
## ⚡ **Quick Start**
### **Option 1: Try the Enterprise Demo** (Recommended)
```bash
cd scaffolding_approach_examples/snowflake_mcp_demo/
# Follow the detailed setup guide in that directory
```
### **Option 2: Try a Simple Script**
```bash
# For math operations (local server)
python3 script_approach_examples/stdio_client_demo.py
# For Cloudflare docs (remote server)
python3 script_approach_examples/sse_client_demo.py
```
---
## 🛠️ **Prerequisites**
- **Python**: Version >= 3.10 < 3.14
- **CrewAI**: Version >= 0.134.0 (for scaffolding approach)
- **API Keys**: Depends on the example (OpenAI, Snowflake, etc.)
- **uv** (recommended) or pip for package management
---
## 📚 **Learn More**
### **Featured Documentation**
- [**🏭 Enterprise Regulatory Intelligence Tutorial**](scaffolding_approach_examples/snowflake_mcp_demo/) - Complete production system
- [**🧮 Mathematician Project**](scaffolding_approach_examples/mathematician_project/) - Local MCP server integration
- [**🔍 Context7 Integration**](scaffolding_approach_examples/crewai_context7_mcp/) - Cloud MCP service
### **Resources**
- [CrewAI Documentation](https://docs.crewai.com/)
- [CrewAI MCP Integration Guide](https://docs.crewai.com/mcp/crewai-mcp-integration/)
- [Model Context Protocol Specification](https://github.com/modelcontextprotocol/specification)
---
## 🤝 **Contributing**
Contributions welcome! Areas where we'd love help:
- Additional MCP server examples
- More transport mechanism demos
- Documentation improvements
- Bug fixes and optimizations
---
## 📄 **License**
This project is licensed under the MIT License - see the [LICENSE.md](LICENSE.md) file for details.
---
## ⭐ **Star History**
[](https://star-history.com/#tonykipkemboi/crewai-mcp-demo&Date)
**Happy building with CrewAI and MCP! 🚀**
