Copilot

You run an MCP server locally to handle multiple micro-centers of processing tasks like customer interviews, E2E testing, and go-live workflows. This setup lets you start specific MCP endpoints quickly, test interactions, and expose dedicated paths for each task area without spinning up separate services manually.

How to use

You interact with MCP servers through an MCP client that communicates with the running server endpoints. To use a specific MCP server, you start the server instance and then access its dedicated path to perform the tasks defined for that server. The project exposes multiple MCP servers, each handling a distinct area such as customers, interviews, or go-live steps. You will typically choose the server you need and then issue appropriate actions that map to that server’s capabilities.

Common usage patterns include selecting a server by name and performing the intended workflow steps. For example, you might start the customer-related MCP and then run interview-related sequences or testing flows against dedicated endpoints. Each server is reachable at its own base URL path, enabling clean separation of concerns and easier testing.

How to install

Prerequisites you need before installing run-time dependencies:

• Python 3.11 or higher

• uvx package manager (used to install and run MCP servers)

Follow these concrete steps to set up the MCP project locally:

# 1) Clone the project repository
git clone <repository-url>
cd CopilotMCP

# 2) Install dependencies using the uvx package manager
uvx install

# 3) Run a specific MCP server by name
uv run main.py --mcp <mcp_server_name>

Additional sections

Configuration and access patterns are designed around named MCP servers. The available servers include hello, customer_mcp, interview_mcp, go_live_mcp, and testing_e2e_mcp. Each server has its own runtime path and can be started independently so you can test one workflow at a time without interference from others.

Running all MCP servers together can be achieved with containerized deployments. Using Docker Compose, you can build and start all services so each server listens on its own port. The exposed endpoints follow a consistent pattern and are ready for local testing and integration with your CI pipelines.

In development environments, you can configure endpoints in your editor to streamline access. For example, you can map a server’s locally hosted path to a friendly URL in your editor so you can trigger actions directly from your editor’s command palette.