enigma-mcp-examples

Minimal examples showing how to use Enigma's MCP tools from an OpenAI Agent (via the openai-agents Python SDK).

Refer to the Enigma MCP Server documentation at https://ai.enigma.com for more details about tools supported, expected results and alternative configurations.

What’s here

• examples/openai-agents/single_agent.py: A simple agent that connects to an MCP server over HTTP and streams a response to a business onboarding question.
• examples/openai-agents/requirements.txt: Python dependencies for the example.

Prerequisites

• Python 3.10+ recommended
• An OpenAI API key (OPENAI_API_KEY)
• Access to an Enigma MCP server endpoint and API key (MCP_SERVER_URL, MCP_API_KEY)

Quick start

1. Create and activate a virtual environment

python3 -m venv .venv && source .venv/bin/activate

2. Install dependencies

pip install -r examples/openai-agents/requirements.txt

3. Configure environment

Create a .env file in the project root (the script uses python-dotenv to load it when run from the root):

OPENAI_API_KEY=your-openai-key
MCP_SERVER_URL=https://mcp.enigma.com/http-key
MCP_API_KEY=your-enigma-api-key

4. Run the example

python examples/openai-agents/single_agent.py

You’ll see the agent stream its reasoning/output to the console.

How it works

examples/openai-agents/single_agent.py:

• Initializes an Agent with instructions to analyze a business.
• Connects to the MCP server over HTTP using MCPServerStreamableHttp with MCP_SERVER_URL and MCP_API_KEY.
• Sends a sample message: “Should I onboard this business? Origin Coffee in Saranac Lake, NY”.
• Streams and prints the model’s output as it arrives.

License

MIT or project’s root license, if applicable.